Objective To evaluate the clinical diagnostic validity of carbon nanoparticle suspension (CNS) in sentinel lymph node biopsy (SLNB) for assessing lymphatic spread of early‐stage cervical cancer. Design A prospective study. Setting and population 356 cases. Methods We enrolled 356 stage Ia2‐IIa2 cervical cancer patients to undergo SLNB using CNS, followed by systematic pelvic lymphadenectomy. All lymph node specimens were assessed using conventional histopathologic ± pathologic ultrastaging analyses. Main outcome measures Sentinel lymph node detection rate (DR), clinical diagnostic validity and various related factors were analysed. Results CNS identified 1456 SLNs in 325 patients. The overall SLN DR was 91.29%. A significantly higher DR was found for patients with tumours <20 mm (97.75% versus 71.91%; P < 0.001). Two patients had false‐negative results. SLNB with CNS had sensitivity of 96.65%, false‐negative rate (FNR) of 4.35% and negative predictive value (NPV) of 99.29%. Importantly, sensitivity (100%), NPV (100%) and FNR (0%) were improved when testing the subgroup of patients with tumours <20 mm (267 cases). There were no observed differences in DR based on pathological type or grade, stage, depth of stromal invasion, surgical approach, menopausal status or prior treatment with chemotherapy (P > 0.05). Conclusions Sentinel lymph node biopsy with CNS results in favourable DR, sensitivity and NPV for women with early‐stage cervical cancer with small tumour sizes. SLNB with CNS is safe, feasible and relatively effective for guiding precise surgical treatment of early‐stage cervical cancer. Tweetable abstract Sentinel lymph node biopsy with carbon nanoparticle suspension is safe and feasible for early‐stage cervical cancer.