Background This analysis was undertaken to assess the need for planned neck dissection in patients with a complete response (CR) of involved nodes after irradiation and to determine the benefit of a neck dissection in those with less than CR by tumor site. Methods Our cohort included 880 patients with T1-4, N1-3M0 squamous cell carcinoma of the oropharynx, larynx, or hypopharynx who received treatment between 1994 and 2004. Survival curves were calculated by the Kaplan-Meier Method, comparisons of rates with the log–rank test and prognostic factors by Cox's proportional hazard model. Results Nodal CR occurred in 377 (43%) patients, of whom 365 patients did not undergo nodal dissection. The 5-year actuarial regional control rate of patients with CR was 92%. Two hundred sixty-eight of the remaining patients (53%) underwent neck dissections. The 5-year actuarial regional control rate for patients without a CR was 84%. Those who had a neck dissection fared better with 5-year actuarial regional control rates of 90% and 76% for those operated and those not operated (p < 0.001). Variables associated with poorer regional control rates included higher T and N stage, non-oropharynx cancers, non-CR, both clinical and pathological. Conclusions With 92% 5-year neck control rate without neck dissection after CR, there is little justification for systematic neck dissection. The addition of a neck dissection resulted in higher neck control after partial response though patients with viable tumor on pathology specimens had poorer outcomes. The identification of that subgroup that benefits from additional treatment remains a challenge.