This article reports the results of an empirical study designed to determine the degree to which college instructors of Spanish can distinguish between machine translation (MT) and non-MT writing samples produced by second language (L2) learners of Spanish in an intermediate-level writing course. We also investigated relationships between detection accuracy rates, instructor teaching experience, and text type (narrative or argumentative), as well as signs instructors consider indicators of both kinds of writing. Results demonstrated that instructors were able to distinguish MT from non-MT writing with a high degree of accuracy by relying on a broad array of indicators. However, neither text type difference nor instructor experience related significantly to detection ability. These findings have practical implications for the L2 classroom with regard to instructor response to MT use and the integration of MT tools to support L2 writing development.