Six extended dyslexic families with at least four affected individuals were genotyped with markers in three chromosomal regions: 6p23-p21.3, 15pter-qter, and 16pter-qter. Five theoretically derived phenotypes were used in the linkage analyses: (1) phonological awareness; (2) phonological decoding; (3) rapid automatized naming; (4) single-word reading; and (5) discrepancy between intelligence and reading performance, an empirically derived, commonly used phenotype. Two-point and multipoint allele-sharing analyses of chromosome 6 markers revealed significant evidence (P < 10(-6)) for linkage of the phonological awareness phenotype to five adjacent markers (D6S109, D6S461, D6S299, D6S464, and D6S306). The least compelling results were obtained with single-word reading. In contrast, with chromosome 15 markers, a LOD score of 3.15 was obtained for marker D15S143 at theta = 0.0 with single-word reading. Multipoint analyses with markers adjacent to D15S143 (D15S126, D15S132, D15S214, and D15S128) were positive, but none reached acceptable significance levels. Chromosome 15 analyses with the phonological awareness phenotype were negative. Parametric and nonparametric linkage analyses with chromosome 16 markers were negative. The most intriguing aspect of the current findings is that two very distinct reading-related phenotypes, reflecting different levels in the hierarchy of reading-related skills, each contributing to different processes, appear to be linked to two different chromosomal regions.