BACKGROUND: Mindfulness-based programmes (MBPs) are increasingly offered at work, often in online self-guided format. However, the evidence on MBPs' effect on work performance (WP) is inconsistent. OBJECTIVE: This pragmatic randomised controlled feasibility trial assessed procedural uncertainties, intervention acceptability and preliminary effect sizes of an MBP on WP, relative to an alternative intervention. METHODS: 241 employees from eight employers were randomised (1:1) to complete a 4-week, self-guided, online MBP or a light physical exercise programme (LE)(active control). Feasibility and acceptability measures were of primary interest. WP at postintervention (PostInt) was the primary outcome for preliminary assessment of effect sizes. Secondary outcomes assessed mental health (MH) and cognitive processes hypothesised to be targeted by the MBP. Outcomes were collected at baseline, PostInt and 12-week follow-up (12wFUP). Prospective trial protocol: NCT04631302. FINDINGS: 87% of randomised participants started the course. Courses had high acceptability. Retention rates were typical for online trials (64% PostInt; 30% 12wFUP). MBP, compared with the LE control, offered negligible benefits for WP (PostInt (d=0.06, 95% CI -0.19 to 0.32); 12wFUP (d=0.02, 95% CI -0.30 to 0.26)). Both interventions improved MH outcomes (ds=-0.40 to 0.58, 95% CI -0.32 to 0.18); between-group differences were small (ds=-0.09 to 0.04, 95% CI -0.15 to 0.17). CONCLUSION: The trial is feasible; interventions are acceptable. Results provide little support for a later phase trial comparing an MBP to a light exercise control. To inform future trials, we summarise procedural challenges. CLINICAL IMPLICATIONS: Results suggest MBPs are unlikely to improve WP relative to light physical exercise. Although the MBP improved MH, other active interventions may be just as efficacious. TRIAL REGISTRATION NUMBER: NCT04631302.