Cigarette smoking is associated with considerable morbidity, mortality, and public health costs. Genetic factors influence both smoking initiation and nicotine dependence, but none of the genes involved have been identified. A genome scan using 451 markers was conducted to identify chromosomal regions linked to nicotine dependence in a collection of 130 families containing 343 genotyped individuals (308 nicotine-dependent) from Christchurch, New Zealand. By pairwise analysis, the best result was with marker D2S1326 which gave a lod score under heterogeneity (H-LOD) of 2.63 (P=0.0012) and a nonparametric linkage (NPL, Zall) score of 2.65 (P=0.0011). To identify regions that warranted further study, rather than comparing the pairwise scores from the scan to theoretical thresholds, we compared them to an empirical baseline, found here to be H-LOD scores of 0.5 and Zall scores of 1.0. We also found a number of large (31-88 cM) regions where many (8-16) consecutive markers yielded small but positive Zall scores. Selected regions of chromosomes 2, 4, 10, 16, 17 and 18 were investigated further by additional genotyping of the Christchurch sample and an independent sample from Richmond, Virginia (91 families with 264 genotyped individuals, 211 nicotine-dependent). Multipoint nonparametric analysis showed the following maximums for the Christchurch sample: Chr. 2 (Zlr=2.61, P=0.005), Chr. 4 (Zlr=1.36, P=0.09), Chr. 10 (Zlr=2.43, P=0.008), Chr. 16 (Zlr=0.85, P=0.19), Chr. 17 (Zlr=1.64, P=0.05), Chr. 18 (Zlr=1.54, P=0.06). Analysis of the Richmond sample showed the following maximums: Chr. 2 (Zlr=1.00, P=0.15), Chr. 4 (Zlr=0.39, P=0.34), Chr. 10 (Zlr=1.21, P=0.11), Chr. 16 (Zlr=1.11, P=0.13), Chr. 17 (Zlr=1.60, P=0.05), Chr. 18 (Zlr=1.33, P=0.09). It is probable that the small samples used here provided only limited power to detect linkage. It may have been difficult therefore to detect genes of small effect, or those that are influencing risk in only a small proportion of the families. When simply judged against the usual standards of linkage significance, none of the individual regions yielded strong evidence in either sample. Some or all of the most positive results in the genome scan of the Christchurch sample, therefore, could be due to chance. However, the presence in the Christchurch scan of multiple large regions containing many consecutive positive markers, coupled with the relatively positive results in these same regions in the Richmond sample, suggests that some of these regions may contain genes influencing nicotine dependence and therefore deserve further study.