Our aim was to check the relative sensitivity of the various commercial kits available on the Italian market for the detection of hepatitis B virus surface antigen (HBsAg) in its two subtypes ay and ad. To this end, panels from different producers were compared by parallel sensitivity assays of several kits. It was found that the kits revealed different sensitivities depending on the panel used, and that sensitivity curves were not parallel. Given the lack of uniformity of response in the various panels currently available, we preferred to use two panels prepared at the Istituto Superiore di Sanità, one for each of the aforementioned subtypes. These HBsAg serial-concentration preparations were used to measure the sensitivity of the different kits on the basis of distinct methodological principles, i.e. RIA, ELISA and RPHA. These are currently considered the most sensitive methodologies and are used for kits classified as belonging to the 3rd generation. For each kit we used the procedure indicated by its manufacturer as being the most sensitive. For some of the kits we adopted all procedures in parallel recommended by respective manufacturers for different conditions of use. Sensitivity was found to be a function of lot, kit, methodology and procedure used. For the RIA and ELISA kits, we also used a modified method in parallel with the normal procedures. This method consisted in presaturating the solid phase in a solution of bovine albumin serum and was generally shown to increase the sensitivity of the kit adopted. For the majority of the kits examined, as well as the direct tests, a confirmatory test was also used, i.e. a specific inhibition test to verify the positivity revealed. In this way an increase in the sensitivity of the sample was also obtained.