Teacher scaffolding, in which teachers support students adaptively or "contingently," is assumed to be effective. Yet, hardly any evidence from classroom studies exists. With the current experimental classroom study we investigated whether scaffolding affects students' achievement, task effort, and appreciation of teacher support, when students work in small groups. We investigated both the effects of support quality (i.e., contingency) and the duration of the independent working time of the groups. Thirty social studies teachers of pre-vocational education and 768 students (age 12-15) participated. All teachers taught a five-lesson project on the European Union and the teachers in the scaffolding condition additionally took part in a scaffolding intervention. Low contingent support was more effective in promoting students' achievement and task effort than high contingent support in situations where independent working time was low (i.e. help was frequent). In situations where independent working time was high (i.e., help was less frequent), high contingent support was more effective than low contingent support in fostering students' achievement (when correcting for students' task effort). In addition, higher levels of contingent support resulted in a higher appreciation of support. Scaffolding, thus, is not unequivocally effective; its effectiveness depends, among other things, on the independent working time of the groups and students' task effort. The present study is one of the first experimental study on scaffolding in an authentic classroom context, including factors that appear to matter in such an authentic context.