Though many educators are familiar with a suggested 3:1 or 4:1 praise-to-reprimand ratio (PRR), relatively little research has been conducted on this subject. Three years of data collected across three states in the United States, from 19 elementary schools and 151 classrooms, during a multi-site efficacy trial were used to analyse the effect of teachers' PRRs on their students' on-task behaviour. Though no PRR threshold (e.g., 3:1, 4:1) was found where behaviour dramatically improved, a positive linear relationship was evident, showing that the higher the teachers' PRR, the higher the students' on-task behaviour percentage. Limitations and implications are discussed. [This paper was published in "Educational Psychology."]