Purpose: Using data from 1,868 children from the US, Australia, and Sweden who took a 10-word spelling test in kindergarten and a standardized spelling test in Grades 1, 2, and (except for the Australian children) Grade 4, we examined two questions. First, does the quality of a child's errors on the kindergarten test help predict later spelling performance even after controlling for the number of correct responses on the kindergarten test? Second, does spelling develop at a faster pace in Swedish than in English? Method: We measured kindergarten error quality based on the number of letter additions, deletions, and substitutions by which each error differed from the correct spelling. Using mixed-model analyses, we examined the relationship of this and other variables to later spelling performance. Results: Kindergarten error quality contributed significantly to the prediction of later spelling performance even after consideration of the number of correct spellings in kindergarten and other relevant variables. The Swedish children showed more rapid growth in spelling than the U.S. and Australian children, a difference that may reflect the greater transparency of sound-to-spelling links in Swedish. Conclusion: Information from a spelling test that is typically discarded--information about the nature of the errors--has value.