OBJECTIVES: To determine whether data obtained from patients returning postal questionnaires accurately reflect how patients receiving imaging-guided lumbar facet injections respond. METHODS: Seventy-eight patients receiving lumbar facet joint injections who returned an outcomes questionnaire (responders) were age and gender matched with 78 patients who did not return the postal questionnaire (non-responders) after facet joint injections. Baseline numerical rating scale (NRS) pain data were collected. NRS and Patients' Global Impression of Change (PGIC) data were collected 1 month after injection by postal questionnaire or telephone interview. Differences in NRS scores were calculated using the unpaired t-test. One level injection patients were compared to patients having ≥2 levels injected using the paired and unpaired t-test. The proportion of patients reporting significant improvement in each group was calculated. RESULTS: NRS scores were significantly improved compared to baseline (p = 0.0001). Thirty-eight percent of responders were significantly improved compared to 50 % of non-responders. Patients having ≥2 levels injected reported significantly higher baseline NRS scores, but by 1 month there was no difference in NRS scores between groups. CONCLUSIONS: Patients returning postal questionnaires report a less favourable outcome. Telephone interview patients having injections at more than one level have better outcomes. MAIN MESSAGES : • Patients returning postal questionnaires report worse outcomes after facet injection. • Method of data collection should be considered when reporting treatment outcomes. • Patients receiving facet injections at more than one level report greater levels of pain reduction.