Aims To investigate the performance of two proposed methods for assessing the prognosis of poorly differentiated thyroid carcinomas (PDTC): the Turin proposal and Hiltzik's histological grade (HHG). This was done using a series of 82 thyroid carcinomas of follicular origin. Results The two methods were able to classify patients accurately into two different prognosis groups. Although the Turin proposal and HHG displayed discrepant cases, they provided similar prognostic information. The Turin proposal gave accurate numbers and thresholds of PTDC criteria (loss of follicular architecture and mitoses, necrosis or convoluted nuclei). One Turin criterion, convoluted nuclei, failed to provide any prognostic value. Hiltzik's histological grade was also a simple and reliable method, allowing detection of tumours with high‐grade features (mitosis and/or tumour necrosis), notably some papillary carcinomas that displayed an intermediate prognosis. We show that Ki67 labelling (≥ 4%) was an independent factor and predictor of cause‐specific survival. Conclusion With similar performances in predicting prognosis, the Turin proposal and HHG provided complementary results in identifying a larger group of ‘intermediate prognosis’ thyroid carcinomas, which require adequate treatment and follow‐up.