Background <p>Coronal pulpotomy (CP) has commonly been used after vital pulp exposure in deeply carious primary molars, although some concerns remain regarding unnecessary pulp removal and procedural morbidity. With the increasing use of calcium silicate cements (CSCs), more conservative vital pulp therapies, such as direct pulp capping (DPC) and partial pulpotomy (PP), have received growing attention. However, direct comparative clinical evidence between these approaches in primary molars is limited. Accordingly, this study sought to compare the cumulative survival probabilities of DPC and PP using RetroMTA^® in deeply carious primary molars with normal pulp or reversible pulpitis.</p> Methods <p>This university-based, assessor-blinded, parallel-group superiority randomized controlled trial enrolled children aged 3–9 years with at least one eligible primary molar. Eighty-eight molars were randomly assigned (1:1) to receive either DPC or PP. All treatments were performed by postgraduate dental students following a strict protocol under the supervision of a single instructor, and all molars were restored immediately with stainless steel crowns (SSC). Follow-up examinations were scheduled approximately every 6 months for up to 24 months. The primary outcome was the cumulative survival probability of pulp treatment, defined by the fulfilment of both clinical and radiographic success criteria. Kaplan–Meier analysis and log-rank tests were used to compare survival probabilities between groups.</p> Results <p>Seventy-eight molars were available for final analysis (40 DPC; 38 PP). Only two radiographic failures occurred in the DPC group, comprising pathological external root resorption at 7 months and furcation radiolucency associated with SSC dislodgement at 11 months. The cumulative survival probabilities were 93.8% for DPC and 100% for PP over the follow-up period (mean: 13.3 months; range: 6–23 months), with no statistically significant difference between groups (<i>p</i> = 0.153).</p> Conclusions <p>PP was not demonstrated to be superior to DPC using RetroMTA^®. Both treatments showed favorable short-term survival under controlled conditions; however, the limited follow-up duration (mean: 13.3 months) and the low number of failure events warrant cautious interpretation. Further adequately powered studies with longer follow-up are required before routine clinical use can be recommended.</p> Trial registration <p>The study was retrospectively registered in the Thai Clinical Trials Registry on 8 May 2024 under the registration number TCTR20240508001.</p>