Some high school seniors who plan to attend college in the fall following graduation do not enroll. This phenomenon, known as summer melt, contributes to lower educational attainment, particularly among low-income students. We extend the literature on summer melt in two ways. First, we show that melt estimates can be sensitive to measurement decisions and local context, and in particular that estimates based on student-level National Student Clearinghouse data can overstate the extent of summer melt. Second, we show that aspects of students' behavior and perceptions, such as their attendance rates, Free Application for Federal Student Aid (FAFSA) completion, and confidence about their intended college, predict summer melt. These results suggest ways that school staff and summer melt interventions could potentially target additional support to college-intending students to facilitate their successful transition to college.