The power of using association analysis to map complex disease genes depends on the strength of linkage disequilibrium (LD) observed in the population and genomic region of interest. In the presence of greater LD, fewer markers need to be typed to identify ones associated with disease. Limited theoretical and experimental data report varying evidence of LD, and additional data are needed to estimate the extent and variability of LD across populations and the genome. We evaluated marker-marker LD using 43 microsatellite markers spanning chromosome 20 with an average density of 2.3 cM. We generated two-marker haplotypes for 837 individuals affected with type 2 diabetes and 386 unaffected spouse controls. All individuals are believed to be of Finnish heritage based on their grandparents' birthplaces within Finland; our sampling scheme under-represents inhabitants of western coastal Finland, an expected source of Swedish admixture. Maximum likelihood estimates were obtained for allele frequencies, by gene counting, and for haplotype frequencies, using the expectation-maximization algorithm. We used a likelihood ratio statistic to test for linkage disequilibrium. A test of homogeneity between cases and controls showed no difference in LD, as expected, allowing the 2446 chromosomes to be analyzed together. Significant (p<0.01) LD was observed using a likelihood ratio test in all (11/11) marker pairs within 1 cM, 78% (25/32) of pairs 1-3 cM apart and 39% (7/18) of pairs 3-4 cM apart. Consistent with chance, significant LD was observed for just over 1% (12/842) of pairs greater than 4 cM apart. Four marker pairs residing on sequence contigs and thus known to be <120 kb apart showed strongly significant LD (p<1x10^-10). These data suggest that microsatellites present at 1-cM density are sufficient to observe chromosome 20 marker-marker LD in a Finnish sample and that whole genome association studies in this population may not require ultra-high marker densities to succeed.