This study investigated the impact of 'assessment for learning' on learner performance in Life Science. Simple random sampling was used to select four schools from the King Cetshwayo District of KwaZulu Natal Province, South Africa, to participate in the study. A quasi-experimental, pretest-posttest comparison group design was used, involving four schools--two forming the 'treatment condition' while the other two served as the 'comparison group'. Altogether, 160 grade eleven learners participated in the study. Two teachers were trained to use assessment for learning (AfL) as an instructional approach, while the two teachers of the comparison group used their usual instructional approaches. Data were analysed using SPSS (V23) and the statistical technique used was the 2-factor ANOVA with repeated measures. The result revealed that learners following an AfL instructional approach performed statistically higher that those following normal classroom instruction. This result is discussed, and recommendations made in respect of both classroom practice and further research. The findings of this study had implications for policy, further research as well as instructional and assessment approaches to be used in the teaching of Life Science in the South African education system.