A study was performed in a rural Mexican community and in Mexico City to investigate possible differences in non-starch polysaccharide (NSP) intakes. One hundred and fourteen women (fifty-five rural and fitty-nine urban) and forty-three men (twenty-four rural and nineteen urban) completed 24 h recall questionnaires for three consecutive days with NSP intakes being estimated from a specially prepared set of new food composition tables. Potential underestimation of intakes was assessed by estimating individual BMR and dividing the estimated energy intakes by BMR to give a ratio. Excluding severe underrecording (ratio < 1·01) suggested NSP intakes of 21·8 and 17·3 g/d in rural men and women and 17·7 and 15·6 g/d in urban men and women respectively. NSP sources differed, with a marked fall in intake from pulses in the urban areas but a compensatory increase of tortilla intakes and a rise in NSP-rich fruit consumption. Both soluble- and insoluble-NSP intakes were higher in rural areas. NSP intakes scaled by probable energy intakes of moderately active adults in developing countries suggest that NSP intakes in Mexico are similar to adjusted values in Africa, i.e. 26·9 g/d in Mexico v. 28·5 g/d in Africa in rural men and 22·9 v. 21·0 g/d respectively in rural women. The NSP: energy density ratio (g/MJ) of the rural diet conformed with the World Health Organization (1990) goal. Data available on urban and rural communities for three countries showed that in each case the intake of NSP was lowest in the urban community.