We describe an extension of Risch's [(1990a,b) Am J Hum Genet 46:222-228, 229-241] method of linkage detection and exclusion for complex genetic traits. The method uses interval mapping to infer disease locus identity-by-descent (IBD) sharing for affected sib pairs (ASPs) based on marker information for the ASP and other genotyped family members. The method is likelihood based, and makes use of Risch's parameterization in terms of recurrence risk ratios for relatives. We describe specific linkage detection and exclusion tests for use as genome screening tools to prioritize genomic regions for further study. We also examine issues of optimal study design. We advocate initially typing a large panel of ASPs (and no additional family members) with a map of genetic markers evenly spaced at 10-20-cM intervals. We recommend a screening procedure that 1) investigates further all regions with maximum lod scores greater than 1 and 2) excludes from consideration those regions that result in lod scores less than -2 at the smallest genetic effect that is viewed as important to detect. Further investigation of an interval might include typing other available families or family members, typing additional markers in the interval, and carrying out further statistical analyses. This strategy is efficient in the number of genotypings required and focuses attention on regions most likely to harbor a disease gene with a substantial impact on disease risk, while resulting in the pursuit of a manageable number of false-positive linkage results. Modification may be required if insufficient ASPs are available or if families come from a significantly admixed population.