The aim was to indirectly validate two short postal questionnaires measuring leisure-time physical activity (LPA) by comparing the answers with serum lipids and anthropometric measurements. Methods All inhabitants aged 31–67 years in two suburban, multicultural areas of Oslo, Norway were invited to “Romsås in Motion”, a community intervention survey, in 2000. Of those, 2950 participants (48%) met and were re-invited in 2003. LPA was measured by two short instruments used since the 1970s (referred to as the Gothenburg instrument) and since 1994 in Cohort of Norway (CONOR). Each instrument was compared with relevant measurements at baseline including LPA according to an adapted version of the long International Physical Activity Questionnaire (IPAQ-L). In addition, changes during 2000–2003 in reported LPA were compared with changes in body weight, waist-to-hip ratio, lipids and other measurements. Results LPA measured by the Gothenburg instrument correlated weakly, but consistently with relevant biological and anthropometrical measurements and with IPAQ-L. The correlation coefficients were −0.1 – −0.2 with triglycerides, total-to-high density lipoprotein (HDL)–cholesterol ratio and waist-to-hip ratio, around 0.1 with HDL–cholesterol and 0.3 with maximal aerobic power. For the CONOR instrument a similar pattern was found in both sexes for the hard LPA and in women for light LPA. LPA measured by each short instrument was in line with LPA measured with IPAQ-L. Conclusions In a multi-linguistic, suburban population in Oslo, Norway, LPA measurements by each of two short questionnaires were weakly, but consistently correlated with relevant biological and anthropometric measurements and strongly with IPAQ-L.