Purpose: To determine the characteristics of ultrasound (US) imaging of completely ablated cases and the effects of duration and clinical experience on accurate microwave ablation (MWA) for the treatment of benign breast tumours. Methods: With written informed consent and approval of the institutional ethics committee, patients with symptomatic or palpable benign breast tumours (longest diameter, 7–32 mm), to whom MWA (2450 MHz) was performed, were enrolled in this prospective nonrandomised study. US and contrast-enhanced US (CEUS) images were applied for follow-up and analysed. Results: Forty-seven consecutive patients with 52 completely ablated tumours were enrolled. Of these 52 tumour ablations in US, 16 ablations were defined as concentric type, and 36 were defined as nonconcentric type. Of these 52 ablations, 7 cases were defined as nonaccurate ablation with the largest margin ≥10 mm in US. The nonaccurate ablation rate in the training group (the first consecutive 30 cases, 7/30) was significant higher than that (the last 22 cases, 0/22) in the practiced group (p = 0.016). Of 38 completely ablated cases (9 mm < the longest diameter <20 mm), the average largest margin in >70 s group was significant larger than that in <70 s group (p = 0.019). Conclusions: Experience was important for accurate MWA in the treatment of benign breast tumour, and at least 30 cases training was recommended. Nevertheless, clinical trials are still required to validate our findings in the future.