Looking through a number of fruit crumble recipes I see that there is a wide range of ratios used for crumble topping. Some use equal amounts of flour (or other dry ingredient), sugar and butter. Others use about half the weight of butter and sugar to flour. Others still use double the amount of sugar and butter to flour.

Is there a definitive ingredient ratio for crumble as there is for pastry (3:2:1)? If so, what is it?

3 Answers 3


There is no single ideal. More sugar and butter will mean a more crunchy crumble top (and one that browns or burns more quickly), more flour will make it more sandy - they each have their place. The juicier the fruit, the sandier I like the topping to absorb some of the juice. The longer the fruit needs to cook, the more resilient to cooking the topping needs to be. And of course for a tart fruit like rhubarb, you might like more sugar in the topping to balance it - or you might just add sugar to the fruit and leave the topping alone.

My goto mix is 1 cup flour, 1 cup sugar, 1/2 cup butter. But play with it a bit and vary it according to the fruit you are using - and I don't just mean "rhubarb" or "apples" but taste a bit of today's fruit to see how sweet it is, how soft it is, etc.

  • Agree, except that I usually add a bit of salt as well. In the US, a stick of butter is 1/2 cup, so a tasty and easy to remember recipe is: 1 cup flour, 1 cup sugar, 1 stick butter, 1 tsp (kosher) salt.
    – Caleb
    Commented Jul 10, 2012 at 14:13

The flour:sugar:butter ratio I have settled on is 1:0.7:0.7

Out of 116 crumble recipes reviewed, I found 58 unique flour, sugar, butter combinations. At the sweetness extremes, there is a recipe with only 6% the weight of sugar to flour, and another with a 1:2.4 ratio. Regarding butter, the ratio range in my sample goes from a lean 1:0.1 through to 1:1.9

Despite an apparent lack of harmony the ratios show, on average, a very clear tendency towards 1:0.7:0.7

I baked a (blueberry) crumble using this ratio and...well actually I burnt the crumble enough to invalidate the experiment. It was still good eating, but I'll need test this ratio again when I'm ready for more crumble. Comment feedback is appreciated from anyone who tries this ratio themselves.

Update: Kate Gregory's excellent answer prompts me to suggest a range of workable ratios based on deviation from the mean:

  • Flour to sugar range: 4:1 to 4:4.6
  • Flour to butter range: 1:0.4 to 1:1

These figures are not backed up with so much as a single experiment and about 37% of the 116 recipes in my sample are outside of these ranges. So take them with a pinch of salt (both literally and figuratively).

  • 1
    I admire your thorough approach, but Kate Gregory's answer suggests that they may not be a single answer. I am curious to see a histogram of your data - maybe you are looking at a bimodal distribution here. Also, have you tried to establish whether one ratio depends on the other one?
    – rumtscho
    Commented Jul 10, 2012 at 11:35
  • You were right to be curious, the histograms are interesting. The flour-to-sugar histo shows peaks at 2:1, 1:1 and 1:1.6 although the last two appear to be noise introduced by the ease of remembering 100g and 1 cup measurements. The butter histo appears to be truly bimodal having peaks at 2:1 and 1:1. Commented Jul 10, 2012 at 12:41
  • To answer rumtscho's last question about dependency, I see that the amount of sugar in a recipe depends more on the amount of butter added than the amount of flour. Here are the correlation coefficients: between flour and sugar 0.43, between flour and butter 0.68 and finally between sugar and butter 0.6. It's possible that butter reduces perceived sweetness and this is why more sugar is added. Commented Jul 11, 2012 at 12:20

Start with the dry ingredients; add the butter one tablespoon at a time till you get the texture you wish?

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.