Objective: Twelve weeks after radiotherapy is the recommended time-point for assessing tumor response in nasopharyngeal carcinoma (NPC); however, regression after 12 weeks remains unclear. We explored NPC regression and the prognosis of patients with delayed clinical complete response (cCR). Materials and methods: MRI images of 556 NPC patients treated with intensity-modulated radiotherapy (IMRT) between 2009 and 2012 were retrospectively reviewed. Clinical tumor response was assessed at 3-4 (assessment 1) and 6-9 months (assessment 2) after IMRT, and survival rates were compared. Results: Of the 556 patients, 463 (83.3%) had cCR at assessment 1 (early cCR). Of the 93 patients with partial response at assessment 1, 45 (48.4%) achieved cCR at assessment 2 (delayed cCR), and 48 did not have cCR at assessment 2 (non-cCR). Locoregional failure rate was lower in patients with a cCR than those without a cCR at assessment 1 (7.1% vs. 26.9%, P<.001) and assessment 2 (7.1% vs. 45.8%, P<.001). Multivariate analysis showed cCR was a favorable prognostic factor for locoregional failure-free survival (LRFFS), failure-free survival (FFS), and overall survival (OS). Early and delayed cCR groups had better 5-year LRFFS (92.6% vs. 93.3% vs. 54.2%), FFS (83.8% vs. 84.4% vs. 48.5%) and OS (92.1% vs. 90.6% vs. 65.4%) than the non-cCR group (all P<.001). Conclusions: Nearly half of the patients with partial response at 3-4 months achieve cCR by 6-9 months; delayed cCR is not a poor prognosticator. We suggest later assessment of cCR at 6-9 months after IMRT is acceptable in responding NPC.