BACKGROUND: Interventions in early childhood education and care (ECEC) services have the potential to improve children's diet at the population level. OBJECTIVE: This study aims to test the efficacy of a mobile health intervention in ECEC services to reduce parent packing of foods high in saturated fat, sugar, and sodium (discretionary foods) in children's (aged 3-6 years) lunch boxes. METHODS: A cluster randomized controlled trial was undertaken with 355 parent and child dyads recruited by phone and in person from 17 ECEC services (8 [47%] intervention and 9 [53%] control services). Parents in the intervention group received a 10-week fully automated program targeting barriers to packing healthy lunch boxes delivered via an existing service communication app. The program included weekly push notifications, within-app messages, and links to further resources, including websites and videos. The control group did not receive any intervention. The primary outcomes were kilojoules from discretionary foods and associated nutrients (saturated fat, free sugars, and sodium) packed in children's lunch boxes. Secondary outcomes included consumption of kilojoules from discretionary foods and related nutrients and the packing and consumption of serves of discretionary foods and core food groups. Photography and weights of foods in children's lunch boxes were recorded by trained researchers before and after the trial to assess primary and secondary outcomes. Outcome assessors were blinded to service allocation. Feasibility, appropriateness, and acceptability were assessed via an ECEC service manager survey and a parent web-based survey. Use of the app was assessed via app analytics. RESULTS: Data on packed lunch box contents were collected for 88.8% (355/400) of consenting children at baseline and 84.3% (337/400) of children after the intervention. There was no significant difference between groups in kilojoule from discretionary foods packed (77.84 kJ, 95% CI -163.49 to 319.18; P=.53) or the other primary or secondary outcomes. The per-protocol analysis, including only data from children of parents who downloaded the app, also did not find any statistically significant change in primary (-1.98 kJ, 95% CI -343.87 to 339.90; P=.86) or secondary outcomes. Approximately 61.8% (102/165) of parents in the intervention group downloaded the app, and the mean service viewing rate of weekly within-app messages was 26% (SD 14.9). Parents who responded to the survey and participating services agreed that it was appropriate to receive lunch box information via the app (40/50, 80% and 6/8, 75%, respectively). CONCLUSIONS: The intervention was unable to demonstrate an impact on kilojoules or associated nutrients from discretionary foods packed in children's lunch boxes. Low app downloads and program message views indicate a need to explore how to improve factors related to implementation before further testing similar mobile health interventions in this setting. TRIAL REGISTRATION: Australian New Zealand Clinical Trials Registry ACTRN12618000133235; https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=374379.