This study aims to validate a frequency questionnaire for categorically ranking people according to recent physical activity (PA) (previous two weeks), applicable to both sexes. Seven one-day PA records were collected for 30 males and 32 females during two weeks. After, the frequency questionnaire was administered covering the same period. The differences (paired-test) between the means obtained by both methods were globally significant for females and for some subgroups of activities in both sexes. The Pearson's correlation coefficient between both methods was 0.78 for males and 0.65 for females. The degree of agreement among methods classifying the subjects on terciles of total PA was lower in males (weighted kappa kw = 0.35, p < 0.06) than in females (KW = 0.52, p < 0.004). Although the study size prevents to infer definitive conclusions, the questionnaire performs better when measuring basic, occupational and total PA than leisure time PA. Its use as a way of ranking individuals in three categories (terciles) of total PA will produce an approximated degree of misclassification of 30% in the lowest tercile of males meanwhile this degree will not go far beyond than 10% for other categories in both sexes.