A study investigated reconstructive recall for linguistic style. It was hypothesized that (1) features of linguistic style would be more difficult to recall than underlying content, (2) reconstructive errors would include stylistic forms recalled as standard forms when subjects lacked productive control of a particular feature of a style, and (3) reconstructive errors would include standard forms recalled as stylistic forms when subjects with productive control of a style attempted to style match. Subjects carried out recall tasks with texts of five different styles: Business, Biblical, Academic, Legal, and Primer. Objective procedures were developed to classify the style of the reconstructed responses, and the results showed that a large proportion of the total responses consisted of the predicted types of reconstructive errors. The reconstructive-style hypothesis was used to integrate a range of experimental findings from studies of memory for text. (Author/FL)