Background: Novel approaches to reduce diabetic foot ulcers (DFU) in low- and middle-income countries are needed. Our objective was to compare incidence of DFUs in the thermometry plus mobile health (mHealth) reminders (intervention) vs. thermometry-only (control). Methods: We conducted a randomized trial enrolling adults with type 2 diabetes mellitus at risk of foot ulcers (risk groups 2 or 3) but without foot ulcers at the time of recruitment, and allocating them to control (instruction to use a liquid crystal-based foot thermometer daily) or intervention (same instruction supplemented with text and voice messages with reminders to use the device and messages to promote foot care) groups, and followed for 18 months. The primary outcome was time to occurrence of DFU. A process evaluation was also conducted. Results: A total of 172 patients (63% women, mean age 61 years) were enrolled; 86 to each study group. More patients enrolled in the intervention arm had a history of previous DFU (66% vs. 48%). Follow-up for the primary endpoint was complete for 158 of 172 participants (92%). Adherence to ≥80% of daily temperature measurements was 87% (103 of 118) among the study participants who returned the logbook. DFU cumulative incidence was 24% (19 of 79) in the intervention arm and 11% (9 of 79) in the control arm. After adjusting for history of foot ulceration and study site, the hazard ratio (HR) for DFU was 1.44 (95% CI 0.65, 3.22). Conclusions: In our study, conducted in a low-income setting, the addition of mHealth to foot thermometry was not effective in reducing foot ulceration. Importantly, there was a higher rate of previous DFU in the intervention group, the adherence to thermometry was high, and the expected rates of DFU used in our sample size calculations were not met.