Abstract Aims and objectives To describe recommendations made by pharmacists in a trial that had found no improvements in selected clinical outcomes (the Seniors Medication Assessment Research Trial, SMART) in terms of expected impact on clinical outcomes and whether they had been implemented by the end of the 5-month period of follow-up. Setting SMART was conducted in a non-academic community practice setting. Method Recommendations made by the pharmacists during SMART, a cluster-randomised controlled trial conducted in family physician offices, were evaluated in this descriptive study. All recommendations to physicians were evaluated independently by two assessors using criteria established a priori and without knowledge of patient outcomes. Each recommendation was evaluated on likely strength and time to impact the patient's health and whether the recommendation was based on published evidence. Relationships between these criteria were analysed. Key findings Overall, the pharmacists made 1099 recommendations for 431 patients randomly assigned to the intervention group or a mean of 2.6 recommendations (standard deviation, 2.1) per patient. A moderate or marked impact on patient health within the 5-month follow-up period would have been expected for 15.5% of all recommendations. At study end, physicians fully implemented 45.8% of the recommendations. Among the recommendations that had been fully implemented, 64.5% of those expected to have a marked impact and 27.5% of those expected to have a moderate impact were anticipated to have the effect on patients' health beyond the 5-month period of follow-up reported in the study results. Conclusion It is likely that one of the contributing factors to not finding statistically significant differences in the SMART study was that only a small proportion of recommendations (15.5%) made by the SMART pharmacists would have an expected clinical effect within the study's follow-up period.