We compared the spelling errors on the WRAT II made by adults (N = 24) with an apparent autosomal dominant form of dyslexia to those made by their normal adult relatives (N = 17) and by spelling-age matched normal controls (N = 17) using a computerized error evaluation program (SEEP). The normal adult relatives were significantly better than the dyslexics in both reading and spelling, but did not differ in age, education, or IQ. SEEP evaluated each error independently for both phonological and orthographic accuracy at 2 levels of complexity. Each level of complexity was analyzed separately using a 3 X 2 (group X dimension) analysis of variance. The main finding of interest was a significant group X dimension interaction effect at the complex level, which indicated that the dyslexics had a qualitatively different profile across the 2 dimensions than either normal group who had parallel profiles. The dyslexics performed like the younger normal group on the complex phonological dimension but like the adult normal group on the complex orthographic dimension. These results indicate a dissociation in this form of familial dyslexia between these 2 dimensions of spelling development, and suggest that these dyslexics may fit the subtype of dysphonetic or phonological dyslexia. The implications of these results for the underlying cognitive deficit in this form of dyslexia are discussed.