Early detection of neck lymph node (LN) recurrence is paramount in improving the prognosis of treated head and neck cancer patients. Ultrasound (US) with US-guided fine needle aspiration (FNA) and core needle biopsy (CNB) have been shown to have great accuracy for LN diagnoses in the untreated neck. However, in the treated neck with fibrosis, their roles are not clarified. Here, we retrospectively review 153 treated head and neck cancer patients who had received US and US-guided FNA/CNB. In multivariate logistic regression analyses, size (short-axis diameter >0.8 cm) (odds ratio (OR) 4.19, P = 0.007), round shape (short/long axis ratio >0.5) (OR 3.44, P = 0.03), heterogeneous internal echo (OR 3.92, P = 0.009) and irregular margin (OR 7.32, P < 0.001) are effective US features in predicting recurrent LNs in the treated neck. However, hypoechogenicity (OR 2.38, P = 0.289) and chaotic/absent vascular pattern (OR 3.04, P = 0.33) are ineffective. US-guided FNA (sensitivity/specificity: 95.24%/97.92%) is effective in the treated neck, though with high non-diagnostic rate (29.69%). US-guided CNB (sensitivity/specificity: 84.62%/100%) is also effective, though with low negative predictive value (62.5%). Overall, US with US-guided FNA/CNB are still effective diagnostic tools for neck nodal recurrence surveillance.