We evaluated two experimenter-delivered, small-group word reading programs among at-risk poor readers in Grade 1 classes of regular elementary schools using a two-arm, dual-site-matched control trial intervention. At-risk poor word readers (n = 201) were allocated to either (a) Direct Mapping and Set-for-Variability (DMSfV) or (b) Current or Best-Practices small-group reading programs, typically for 10-11 hr over 10 weeks. Students were matched on baseline reading and language abilities, parent demographic measures, and observed regular classroom teaching quality. Results of hierarchical data modeling showed advantages for the DMSfV program (p < 0.05 for word reading and spelling at posttest and word reading and sentence comprehension at 5-month delayed posttest), with discernible valued added for the DMSfV condition across all follow-up measures. Results support the use of small-group preventative literacy intervention models that teach both direct mapping of grapheme--phoneme correspondences in text and set-for-variability.