Purpose Aims of this study were to evaluate the agreement between the short and long versions of the International Physical Activity Questionnaire (IPAQ: Italian versions), their reproducibility (agreement and reliability) and construct validity (relative to pedometry) in a clinical population. Methods Ninety patients affected by obesity ( N  = 39), type 2 diabetes mellitus ( N  = 26) or both ( N  = 25) were recruited. They were asked to maintain their usual physical activity habits during two consecutive weeks and to fill the questionnaires twice (at the end of each week). They were also asked to wear a pedometer for 7 consecutive days after the first administration of the questionnaires. Results We found acceptable agreement between the IPAQ short and long versions (ICC 2,1 values were 0.81 and 0.77 for the 1st and 2nd administration), uncertain reproducibility (acceptable reliability but poor agreement) and inadequate validity relative to pedometry (the correlation coefficients between all IPAQ scores and daily steps were <0.50) for both IPAQ short and IPAQ long. Conclusions The IPAQ use may be justified in daily clinical practice and in clinical research (e.g., in cross-sectional studies) for a simple and rapid evaluation of the physical activity level for discriminative purposes. However, the use of these questionnaires does not appear suitable for prospective interventional studies in which the level of physical activity of the recruited patients has to be assessed over time.