The relationship between reading skill and syntactic-processing efficiency was investigated in this experiment. Adults were classified into four groups that varied independently as a function of comprehension level and reading speed. They then performed three sentence-processing tasks—meaning classification, grammaticality classification, and single-sentence question-answering. In each task, they read sentences containing temporary structural ambiguities, the alternative interpretations of which either did or did not conform to the principle of‘minimal attachment’ (cf., Frazier and Fodor, 1978; Frazier, 1979). Three sentence types were included: transitives, normal complements (with ‘that’), and reduced complements (without ‘that’). Evidence of garden-pathing was found in the grammaticality-classification and question-answering tasks, but not in the meaning-classification task. However, these effects did not generalize to all reader groups. Good fast and average slow readers (arguably the best and worst readers respectively) showed very little evidence of garden-pathing in any task.