We tested whether urban middle-school students from mostly low-income homes had improved academic vocabulary when they participated in a freely available vocabulary program, Word Generation (WG). To understand how this program may support students at risk for long-term reading difficulty, we examined treatment interactions with baseline achievement on a state standardized test and also differential effects for students with (n = 398) and without (n = 1,395) individualized education plans (IEPs). Students in this unmatched quasi-experiment (5 WG and 4 comparison schools) completed pre- and postvocabulary assessments during the intervention year. We also retested student vocabulary knowledge after summer vacation and the following spring on 11 target words to construct a longitudinally consistent scaled score across 4 waves of data. Growth models show that students experienced summer setback. Although there were no average underlying differences in growth or differences in summer setback for students by baseline achievement, better readers improved more from program participation. IEP status did not predict differential benefits of program participation, and students with IEPs maintained gains associated with participation in WG; however, participation in the program did not change underlying growth trajectories favoring students who did not have IEPs.