Summary.This paper reports the continuation of an earlier investigation into selection for remedial education (Lytton, 1961). The children's gains over the whole of the remedial period were recorded and each child was followed up one to two years after he left the remedial group. While children lost some of their original gains in reading as well as in arithmetic during the follow‐up period, the losses in arithmetic were much greater. Judged by their standing in relation to their age groups (Achievement Ratio), some children maintained a large part of their gains after remedial teaching ceased, especially in reading.In reading, selection by tests and selection by teachers' judgment did not produce significantly different long‐term results. In arithmetic, higher gains were associated with test selection. Both in reading and arithmetic, children with high I.Qs. recorded greater gains during the remedial period and maintained them better afterwards than children with a low I.Q. However, in reading‐as opposed to arithmetic—even children of below average intelligence maintained some of their gains. A.Q. was an unsatisfactory predictor of success. Those who had changed to secondary schools during the follow‐up period lost more than those who were still in primary schools. Maladjustment was found to be inversely related to long‐term gains due to remedia