Archival data were analyzed with piecewise growth models to determine the seasonal growth of students not reading proficiently who did (treatment students = 144) and did not (control students = 1113) participate in their school district's summer reading program. The rising first- through fifth graders (48% female) were predominately White (74%) and economically disadvantaged (79%). Control students tended to exhibit stable or declining spring-to-fall performance that was significant only in Grade 2, where a measurement artifact may have influenced results. When including a summer testing point in the models for treatment students, performance tended to be stable or declining while the students were receiving summer instruction but then improved during their subsequent summer-to-fall break. Conditional models revealed demographic subgroups had varying patterns of higher or lower growth rates compared to the reference groups, but rarely were these significant. Issues with measurement and implications for planning summer programs are discussed.