Cascade models of in-service training are widely considered to be a cost effective means of introducing educational change to large numbers of teachers. Data from 511 teachers completing a cascade training programme that introduced current ideas about and procedures for teaching English to young learners, suggests that provision of training alone is no guarantee that cascade training aims will actually be applied in classrooms. The paper considers implications for cascade projects, suggesting that planning needs to be a parallel process if an adequate return on outlay, in the sense of teachers applying skills introduced in training in their classrooms, is to be achieved.