Prior studies with anthropomorphic phantoms and single, static in vivo brain images have demonstrated that scatter correction significantly improves the accuracy of regional quantitation of single-photon emission tomography (SPET) brain images. Since the regional distribution of activity changes following a bolus injection of a typical neuroreceptor ligand, we examined the effect of scatter correction on the compartmental modeling of serial dynamic images of striatal and extrastriatal dopamine D(2) receptors using [(123)I]epidepride. Eight healthy human subjects [age 30+/-8 (range 22-46) years] participated in a study with a bolus injection of 373+/-12 (354-389) MBq [(123)I]epidepride and data acquisition over a period of 14 h. A transmission scan was obtained in each study for attenuation and scatter correction. Distribution volumes were calculated by means of compartmental nonlinear least-squares analysis using metabolite-corrected arterial input function and brain data processed with scatter correction using narrow-beam geometry micro (SC) and without scatter correction using broad-beam micro (NoSC). Effects of SC were markedly different among brain regions. SC increased activities in the putamen and thalamus after 1-1.5 h while it decreased activity during the entire experiment in the temporal cortex and cerebellum. Compared with NoSC, SC significantly increased specific distribution volume in the putamen (58%, P=0.0001) and thalamus (23%, P=0.0297). Compared with NoSC, SC made regional distribution of the specific distribution volume closer to that of [(18)F]fallypride. It is concluded that SC is required for accurate quantification of distribution volumes of receptor ligands in SPET studies.