Although the learning style construct has aroused much interest over the years, questions remain regarding basic issues such as definition, the validity and/or reliability of various measurement instruments, and the relationship between learning style and successful learning. Furthermore, although maintaining stylistic flexibility is recommended by many authors, few studies have attempted to relate the style-stretching concept to successful learning. This study therefore attempted to address these questions. According to results, conducted among 106 Turkish university students, using an original instrument constructed using elements from established questionnaires, a small group of styles was significantly correlated with exam results, accounting for about a quarter of the variance (considered a large effect size in social science). In addition, higher-scoring students reported a more eclectic range of styles, suggesting more willingness to style-stretch, while lower-scoring students reported a more limited range. Pedagogical implications as well as areas for ongoing research are suggested.