QUESTION: Does the addition of telephone coaching to usual physiotherapy care improve activity for people with non-chronic low back pain and low to moderate recovery expectations? DESIGN: Randomised trial with concealed allocation and intention-to-treat analysis. PARTICIPANTS: People attending the physiotherapy department of a public hospital for treatment within eight weeks of onset of non-specific low back pain. Eligible participants had low to moderate recovery expectations, defined as a response of 7 or less to the question 'How certain are you that you will return to all of your usual activities one month from today?' on a scale from 0 (not certain at all) to 10 (completely certain). INTERVENTION: Five sessions of telephone coaching by a physiotherapist trained in health coaching techniques in addition to usual physiotherapy compared to usual physiotherapy alone. OUTCOME MEASURES: The Patient Specific Functional Scale, Oswestry Disability Index, Pain Self Efficacy Questionnaire, and recovery expectation were measured at baseline, 4, and 12 weeks. RESULTS: 30 participants were recruited, with 26 completing all measures at 12 weeks. There were no significant differences between groups at 4 weeks. After 12 weeks the coaching group improved significantly more than the control group on two 10-point scales: the Patient Specific Functional Scale (mean difference 3.0 points, 95% CI 0.7 to 5.4) and recovery expectation (mean difference 3.4 points, 95% CI 1.1 to 5.7). Estimates of effect sizes were moderate to large in favour of the intervention. CONCLUSION: The addition of telephone health coaching to usual physiotherapy care for people with non-chronic non-specific low back pain led to clinically important improvements in activity and recovery expectation. TRIAL REGISTRATION: ACTRN12607000458437.