Despite the promotion of data-driven or data-informed instructional practices in teacher education and professional development, past research indicates that teachers use a limited number of sources for student data to make short-term adjustments to their teaching in order to address deficiencies in student learning. Science teachers, with a more robust background in data collection and analysis as well as expectations to teach data practices to students, conceivably, might be more comfortable in implementing data-driven instructional practices. This mixed-methods study explores firstly, how secondary science teachers in the USA define "data-driven instruction" and secondly, how they report using student data in their instructional practice. Responses from a nationally disseminated mixed-methods questionnaire (N = 451) were analyzed. Overall, secondary science teachers limit their definition of data to student assessments and use that information to inform their daily instructional practice rather than improve their teaching overall. Consistent with previous findings, this study's findings also provide evidence for hope of improvement in teachers' data practice. In addition, there was a small, but noticeable, group of secondary science teachers who appeared to be jaded with or to misunderstand the effective teaching skill conveyed by the term data-driven instruction. Implications for teacher education and professional development are discussed.