In three experiments, we compared the effectiveness of rainbow writing and retrieval practice, two common methods of spelling instruction. In experiment 1 (n = 14), second graders completed 2 days of spelling practice, followed by spelling tests 1 day and 5 weeks later. A repeated measures analysis of variance demonstrated that spelling accuracy for words trained with retrieval practice was higher than for words trained with rainbow writing on both tests (?[subscript p][superscript 2] = 0.49). In experiments 2 (second graders, n = 16) and 3 (first graders, n = 12), students completed 2 days of spelling practice followed by a spelling test 1 day later. Results replicated experiment 1; spelling accuracy was higher for words trained with retrieval practice compared with rainbow writing (?[subscript p][superscript 2] = 0.42 and 0.64, respectively). Furthermore, students endorsed both liking and learning from retrieval practice at least as much as (and sometimes more than) rainbow writing. Results demonstrate that retrieval practice is a more useful (and as engaging) training method than is rainbow writing and extend the well-established testing effect to beginning spellers.