We tested effects of dissuading students from requesting help unless they really needed it. The manipulation occurred while the students solved problems on an ITS that provided proactive help. Compared to their counterparts, dissuaded students requested help much less often. Moreover, the less help students requested, the higher their posttest score. Among students with lower pretest scores, dissuaded students marginally gained more than their non-dissuaded counterparts. We discuss our results, a new type of help abuse, some ramifications of proactive help, and the generalizability of our results.