web analytics
≡ Menu

Experiments in Dulce de Leche 2 – A Question of Time

Well over a year ago, a neophyte food blogger once pondered the question of whether or not the specific brand of sweetened condensed milk would significantly affect the quality & consistency of the dulce de leche it would later produce.

Other questions then arose. Could dulce de leche be produced from a tube of sweetened condensed milk? Does coconut cream, under the same circumstances, produce an equivalent product using the same methods? The answers to these questions were explored and published in a previous post.

However, there was one question that came to mind through the course of the experimentation that was left unanswered. It was a question of time.

For those of you that are unfamiliar with the procedure of making dulce de leche, that sublime milk caramel of Latin American origins, there are two broad methodologies.

The original involves the combination of milk & sugar, which is either stirred in a saucepan or baked in the oven until it reached the desired consistency. The other involves simmering a tin of sweetened condensed milk in a pot of water. Far less labour intensive but if you’re not careful, it comes with some perils of its own beyond burnt caramel.

The thing about producing dulce de leche using sweetened condensed milk as the base is that most recipes recommend a range of usually anywhere between 1 – 3 hours cooking time (I’ve seen one that goes up to 8 hours!) However, which cooking time(s) produces optimal results? Why is there such a variance in time? How different are the results between the 1 hour and 3 hour versions?

During the previous experiment time remained a constant, 3 hours cooking time, whilst the brands and contents varied. This time around the brand will remain constant whilst the time will be varied, in order to see what impact time will have on the end result.

The process of the experiment follows, for the most part, the standard procedure for producing dulce de leche using sweetened condensed milk. Namely, place an unopened tin of sweetened condensed milk into a pot, cover with water, simmer for a period of time and voila! Milky caramel heaven.

The experiment deviates a little from this procedure, in an attempt to apply a degree of consistency with the end results. The procedure for the experiment is as follows:

  1. All tins were placed into an empty pot, covered with cold water and then brought to the boil.
  2. Once boiling point was reached, the water was turned down to a simmer and then a timer was started to go off after a 30 minute interval.
  3. At each interval, one tin was removed, whilst the remaining tins were rearranged to be evenly distributed in the pot.
  4. The pot is topped off with boiling water from a kettle, brought to the boil once more, lowered to a simmer and then the timer was started for the next 30 minute interval.
  5. For the tins that were removed, they were shocked in cold running water to cool then down as quickly as possible in order to halt the cooking process.
  6. Once all tins were removed and cooled, they were left still sealed on a bench overnight to ensure that they all were all at room temperature the following day.

Please don’t mistake the above procedure as following the scientific method in any way, as I have no illusions that this is likely not the case. It’s simply an attempt on my part to minimise the impact of the variables in the experiment so that when and if you attempt to use the information from this post at home, that your results, though perhaps not exact, will at least be in the same ballpark.

Below is the breakdown of the various characteristics of dulce de leche produced during a span of three and a half hours.

In honor of its Latin origins, labels are assigned to each variety in Spanish, or at least what passes as such with Google Translate as I know little of the language beyond what I might have learned from Looney Toons cartoons.

The labels are there mainly as a short hand reference to use within the post, as I’m not aware of an equivalent grading system in existence. It was also fun to make up distinct names for each, so that may have something to do with it as well.

Apologies to any Latinas that may be offended by my poor use of Spanish.

Grades of Dulce de Leche

  • 0.0 hrs crudo (raw) aka sweetened condensed milk
    • Phase: Fluid
    • Flavour: Milky
    • Consistency: Runny
    • Usage: Teh Tarik
    • Notes: Sweetened condensed milk in its uncooked unaltered state. Included as a point of comparison.
  • 0.5 hrs – semicrudo (undercooked, semi raw)
    • Phase: Fluid
    • Flavour: Retains milky qualities but with a more mellow sweetness.
    • Consistency: Treacle/molasses like
    • Usage: As a light jam spread or for very delicate items.
    • Notes: Yet to adopt distinct caramel characteristics at this stage. Just a more mellow sweetness than crudo, as stated.
  • 1.0 hr – blanco (white)
    • Phase: Solid
    • Flavour: Still distinctly milky but with some caramel notes. Similar in flavour to butterscotch.
    • Consistency: Similar to margarine.
    • Usage: As a filling, sandwiched between layers of light sponge cake, when a soft texture is required.
    • Notes: First solid state at room temperature.
  • 1.5 hrs – rubio (blonde)
    • Phase: Solid
    • Flavour: Caramel qualities take over milky ones as dominant flavour.
    • Consistency:  Peanut butter like.
    • Usage: For shortbread cookies or other semi-firm biscuit applications.
    • Notes: First stage to taste like “dulce de leche”.
  • 2.0 hrs – pajizo (straw)
    • Phase: Solid
    • Flavour: Loses the “raw” milky quality that resembles the crudo state.
    • Consistency: Resembles soft chewy caramel in texture.
    • Usage: Similar to rubio but where a firmer texture or more pronounced caramel flavour is desired.
  • 2.5 hrs – marrón (brown)
    • Phase: Solid
    • Flavour: First signs of a subtle bitterness in the caramel flavour.
    • Consistency: Marginally firmer & stickier than pajizo.
    • Usage: Sandwiched between firm biscuits, or in applications where the slight bitterness is desired.
  • 3.0 hrs – marrón oscuro (dark brown)
    • Phase: Solid
    • Flavour: Bitterness comes through a little more than marrón. Resembles bitter qualities of a semi-sweet dark chocolate.
    • Consistency: Like cold cream cheese.
    • Usage: Same as marrón but where stronger flavours and firmer texture is desired.
    • Notes: Mouth feel is thick & tacky, verging on being unpleasant in texture.
  • 3.5 hrs – negro (black)
    • Phase: Solid
    • Flavour: Strong caramel flavour with pronounced bitterness, but not entirely unpleasant.
    • Consistency: Putty-like. Very thick, unpleasant mouth feel.
    • Usage: Ah, maintaining structural integrity of buildings?

When it comes to negro, I’m only half-joking about its usage in the structural integrity of buildings. It’s stickiness and well as its thick, putty-like qualities give me the impression that this may work, in sufficient amounts, as an alternative to royal icing when building gingerbread houses. Though this has yet to be tested, you’ll see what I mean in a moment.

For a better representation of the consistency and viscosity of the various grades of dulce de leche, refer to the following video which explores the spreadability of each variety.

The sound track to the video is Garden of the Forking by J.Lang, licensed under Creative Commons.

Now, some of you may want to know which grade produces the “best” dulce de leche. For me, the question is a matter of usage.

If you want to have dulce de leche as a table spread, especially for something crumbly like scones for example, then I would go either semicrudo or blanco (0.5 to 1 hr). If the application is for a sponge cake, I’d likely go either semicrudo or blanco again (0.5 to 1 hr) if its to be applied as a spread. If using a piping bag, I wouldn’t go beyond pajizo (2 hrs), mostly due to the texture & consistency. As a filling for firm biscuits, then I’d probably go with at least a pajizo or marrón (2 to 2.5 hrs) to minimise the filling from being squished out from the biscuit sandwich when bitten into.

Though technically I wouldn’t consider anything below rubio (1.5 hrs) representative of the “true” dulce de leche flavour, in my eyes the application best dictates the appropriate variety of dulce de leche to employ. Having said that, my personal favourites are blanco, rubio & pajizo i.e. 1 – 2 hour cooking time.

One thing to note, if some of this wasn’t apparent in the video, is that as the dulce de leche gets progressively darker, the consistency becomes thicker & takes longer to dissolve in the mouth. The effect is much like having a large mouthful of peanut butter. So if you have a biscuit that on its own would “dry” out the mouth, requiring the services of a beverage as an accompaniment, then you may not want to lather on a thick layer of some of the darker versions of dulce de leche. Either go with a lighter variety or use it sparingly.

A final note. Regarding the specific brand selected for this experiment, this was selected intentionally, not due to any sponsorship or endorsement reasons, but simply due to its wide available in Australia. If you use a different brand, your results may differ. Though, hopefully they’ll fall somewhere within the grading system listed above, so you’ll be able to adjust relative cooking times to reach the desired consistency & flavour.

There are still more questions to explore regarding this simple-to-make yet delectable confection. Something that I’ll likely look into at some stage. Perhaps in a year. Perhaps sooner. Who knows…

Dulce de Leche (Rubio)


One or more tin of sweetened condensed milk


  1. Place one or more tins of sweetened condensed milk unopened in a pot and cover with cold water so that it covers the top of the tin.
  2. Bring the pot to a boil and then lower to a simmer.
  3. Simmer for 90 minutes. Top up with water if the top of the tin is no longer submerged.
  4. After 90 minutes, remove the tin from the pot and immediately cool by placing it into a pot with cold running water until the tin is cool to the touch.
  5. Once the dulce de leche has sufficiently cooled, open the tin and use as a filling for cakes or banoffee pie, soft biscuits, topping on vanilla ice cream etc.


  • Any of other grades of dulce de leche can be made using the above recipe. Adjust the simmering time according (Step 3) to reach the desired consistency.
  • Optionally, place a small tea towel at the bottom of the pot to prevent the tins from making a rattling sound as it simmers.
  • Do not leave the pot unattended for an extensive period of time. If all the water boils out of the pot, enough heat could build up within the tin to make it explode causing, at best a mess to clean up, or at worst property damage and/or personal injury.
  • The reason for topping up the water above the top of the tin is for consistent cooking throughout of the tin. From my experience, so long as there is sufficient water in the pot, the tin should be safe from explosions.
  • The running water doesn’t need to run at full stream. A gentle trickle, so long as there is movement in the water, will go a long way to cooling the dulce de leche than still water alone.
Related Posts with Thumbnails
{ 32 comments… add one }
  • Chris January 14, 2011, 1:37 pm

    I take my cap off to you sir. Good experimental design, and very useful information. Now, to muster the courage to make some for myself – the idea of exploding caramel and resultant shrapnel and sticky mess still scare me

    • Simon January 15, 2011, 3:28 am


      Look at it this way. I had seven tins going at once and lived to tell the tale (as long winded as it was). One should be a push over so long as you’re mindful of water levels and keep an eye on it.

      Oh, and you may want to follow the advice about placing a tea towel at the bottom of the pot. The rattling can be unnerving for a first timer.

  • Rhonda (The Dainty Baker) January 14, 2011, 1:56 pm

    ohhh gosh how very interesting!!! love the mesh of science and food

    • Simon January 15, 2011, 3:29 am

      So do I! Though, for me it was more about getting consistent results rather than being specifically scientific about it.

  • Gaby January 14, 2011, 1:59 pm

    Hey Simon, great experiment! I’m always delected to hear how non-latinos enjoy our dulce de leche. Only one small correction to your labels: because dulce de leche is “male”, it should be “rubio” instead of “rubia”.

    • Simon January 15, 2011, 3:29 am

      Thanks for the correction! Updated the post accordingly :)

  • OohLookBel January 14, 2011, 2:08 pm

    Well done, Simon, this is an excellent report with first-rate analysis. You should submit it to the International Council of Breakfast Spreads (ICBS).

    • Simon January 15, 2011, 3:29 am

      International Council of Breakfast Spreads? You’re teasing me, aren’t you? :)

  • Susan January 14, 2011, 2:58 pm

    How very interesting! I have never made this caramel before, but have been wanting too for ages. So it’s good to know how long I am looking at boiling the can.

    • Simon January 15, 2011, 3:31 am

      It’s worth the very tiny amount of effort required :)

  • KayB January 14, 2011, 3:45 pm

    My grandmother always taught me ‘never more than 2 hours’. Oh, how the oldies always know best! Thanks so much for sharing, this will really come in handy.

    • Simon January 15, 2011, 3:31 am

      They know so much, don’t they? :)

  • Sara @ Belly Rumbles January 14, 2011, 6:35 pm

    Simon what an awesome post, very informative and interesting. Have wondered about time factors and love the comparison over the 3 1/2 hours.

    I had to have a giggle too at the end of your spreading demo, though I almost felt sorry for the piece of bread being spread with the negro.

    • Simon January 15, 2011, 3:34 am

      Lol! Yeah, it was annoying but it worked out well as a way to end the clip.

      Now we needn’t wonder no longer! However, there are still a few questions I wouldn’t mind finding out the answer to. In due time I guess :)

  • Melissa January 14, 2011, 6:38 pm

    wow! that’s a really great approach to know the best way to produce the best caramel. oh goodness i should try making this too!

    • Simon January 15, 2011, 3:36 am

      Please do! It takes almost no effort and the results are so nice, let alone their derivative applications :)

  • Moya January 14, 2011, 7:15 pm

    What a beautiful infographic your dulces de leche err dulce de leches…ahhh what a beautiful infographic your caramels make! Congratulations on some fine research sir!

    • Simon January 15, 2011, 3:38 am

      Thanks! Much appreciated :)

  • yewenyi January 15, 2011, 8:18 am

    Excellent test there Simon, perhaps you could do the same say for Beef Rendang, I am never sure how long to cook it for.

    Also, some enterprising company could sell precooked cans of a particular amount of time to save all the effort.

    • Simon January 15, 2011, 1:23 pm

      I don’t think I could do the same thing with beef rendang. First, you can taste it as you go (unless a pressure cooker is involved). Secondly, there are too many variable involved – the composition of the sauce, how thick the meat is cut, the shape of the meat pieces, the type of meat used, the specific breed of animal, and the list goes on.

      Companies do. The very brand referenced in the post have them available pre-made. However, you’d have to live with whatever consistency they give you and it costs more too.

  • yewenyi January 15, 2011, 8:19 am

    Of course, now you might have to do the multiple line-up test with your favourite 1.5 cooking time.

    • Simon January 15, 2011, 1:23 pm

      I’m not sure what you’re getting at with regards to multiple line ups.

  • mademoiselle délicieuse January 15, 2011, 8:50 am

    Haha, love how the darkest version holds its integrity as a little ball almost refusing to be spread at all on the bread!

    And love the rainbow of dulce de leche =)

    • Simon January 15, 2011, 1:25 pm

      Tough little bugger :P

      As I mentioned in the post, the issue with the last one was more of a problem with texture and not as much taste.

  • penny aka jeroxie January 15, 2011, 4:02 pm

    Gosh! Great experiment… love this post. :)

  • Kate January 24, 2011, 8:45 am

    AWESOME POST! I have made the ‘3 milks’ cake before after drooling over it in many southern American blogs but only used condensed milk straight out of the can as I had no idea really what dulce de leche was!

  • Kate January 24, 2011, 8:48 am

    Okay, another comment; I just went back and watched the video.




    • Simon January 24, 2011, 11:44 pm

      Ah, the tres leche cake. Something that I’d love to try and make at some stage.

      Thanks for the comment. If you intend to make your own dulce de leche, hope you enjoy it :)

  • Dimitrios Dalagiorgos January 24, 2016, 4:03 am

    I’ve left the condensed milk to boil for at least 3 hours. Then stored it for several months. When I opened it its consistency was nowhere near the consistency shown in the video; mine was too runny compared to that. So, what do you suppose happened? It didn’t cook as it should or somehow it desolved over time became fluid again?

    • Simon January 28, 2016, 3:34 pm

      Hi Dimitrios :)

      I’m not sure that I can give you an answer as to what may have been the cause. However, what I can say is that I’ve noticed differences in consistency between brands as well as some sweetened condenses milks being made up of different ingredients, either with varying ratios of sugar and milk or also including thickeners as well, which can affect the texture. It may have been the long storage after cooking but having never tried that (and good luck to it lasting that long in storage in my household!), I really cannot say.

      Sorry that I can’t be of more help but perhaps try and give it a go again and check out the consistency when the contents are fresh. See how you go with that.

Leave a Comment

CommentLuv badge

Next post:

Previous post: