Aims: This study investigated whether the number of words provided in naming therapy affects the outcome. A second aim was to investigate whether severity of anomia should be used to determine the number of words provided in therapy. Methods & Procedures: First a meta-analysis of 21 anomia treatment studies between 1985 and 2006, yielding 109 individual datasets, explored whether the number of items provided and the severity of anomia influenced the success of therapy. The second part was a cross-over case-series study with 13 individuals with aphasia who had varying degrees of anomia. Individuals received two blocks of therapy (each of 10 sessions) where the set size of items to be learned was manipulated: either a small (n = 20) or large (n = 60) set in each block. Therapy and control sets were matched for baseline naming ability, frequency, phoneme, and syllable length. Therapy consisted of progressive phone- mic and orthographic cues until successful naming was achieved. All word sets (small, large and control) were retested immediately after each thrapy block finished (within 1 week) and 5 weeks after the end of each block of therapy. Outcomes & Results: The meta-analysis showed a large variation in the number of items given to participants to learn (from 5 to 120 items) and very different learning outcomes that were not linked to the number of items given. The current literature contained an unexpected bias in that, across studies, more items were given to those with severe aphasia. Consequently, the meta-analysis could not provide a clear answer to how may items should be given in therapy-thus motivating a direct comparison in a new therapy study. We found significant gains in naming accuracy for both the small (n = 20) and large (n = 60) therapy sets immediately and 5 weeks post therapy. Proportionally, there was no difference between the two sets for the group as a whole, although there was individual variation in the overall therapy effect. If expressed as the raw numbers of words learned after therapy, this means that 12 of the 13 participants learned more words when given the large (n = 60) set. Severity of anomia correlated with learning performance but did not interact with set size. Conclusions: The empicial study suggested that people with anomia could tolerate more items in therapy and that the severity of anomia should not necessarily determine how many words should be given in therapy. © 2010 Psychology Press.