Background: Most cases of low back pain (LBP) are regarded as non-specific and current studies indicate that for many this is a chronic recurrent condition, in which people experience episodes of pain with symptom-free periods in between. It is likely that acute exposure to some factors triggers the reappearance of new episodes in recurrent LBP regardless of the causality of the underlying condition (i.e. risk factors). Additionally, it has been shown that LBP patients present with different trajectories and different trajectories possibly have different triggers. Hence, dividing patients into some clinically meaningful subgroups may offer new insights into triggers, effective preventive strategies and, therefore, prognosis. This study aims to identify self-reported triggers and trajectories of episodes of recurrent LBP and to examine the prognostic association between different triggers and LBP trajectories. Methods: This is a longitudinal, multicentre, Australia-wide observational study of patients with recurrent non-specific LBP. Two hundred adults with at least a one-year history of LBP will be recruited from primary care clinics or private practices and followed for a year. Each will receive an SMS every fortnight (26 time-points in total) enquiring the occurrence of a new episode of pain in the past 2 weeks and its intensity. Upon report of a new episode, a telephone interview will be performed to appraise exposure to self-nominated triggers in a period of 24 h preceding the pain. Trajectories will be identified by latent class analysis at the end of the follow-up based on the pain intensity, frequency, and length of episodes. Triggers will be categorised into physical and psychosocial groups. Generalised linear mixed models with logit links will be used to explore pain triggers associated with pain trajectories. Discussion: The completion of this study will provide insight into the patients' self-reported triggers of LBP and also their possible prognostic association with different trajectories. Some newly-identified and pre-identified triggers are likely to be found and reported.