The diagnosis of well differentiated carcinoma (i.e papillary carcinoma and follicular carcinoma) represents one of the most challenging issue in thyroid pathology. Aim of the present review is to discuss new perspective and old problems in this topic. Three main subjects are developed, corresponding to: 1) the role of fine needle aspiration versus frozen section examination in pre- or peri- operative diagnosis; 2) the management of small papillary tumour; 3) pathological classification of those tumours indeterminate for papillary or follicular nature. There is general agreement that fine needle aspiration represent the best pre-operative diagnostic tool for thyroid nodules; foremost limits are represented by "not diagnostic" and 'follicular lesion, NOS". The former should be repeated or, if suspicious for papillary lesion, improved with intra-operative apposition cytology; the latter should be deferred to histology with frozen section evaluation reserved to those institution with daily practice on this issue. The management of papillary micro-carcinoma (i.e. papillary carcinoma smaller than 1 cm.) in the setting of an otherwise benign thyroid disease is a matter of debate, since several clinicians suggest to consider these as incidental findings thus avoiding additional treatment. Recently this attitude has been supported by the proposal to regard these lesion as "tumour" and not carcinoma: available data on follow up seems to sustain and favour this approach. There exist a group of well differentiated tumours of the thyroid lacking the criteria to be diagnosed either as papillary (i.e. nuclear grooves, nuclear pseudo-inclusion and nuclear clearing) or follicular (i.e. capsular or vascular invasion) carcinoma; for these lesion, whose behaviour (nodal or blood metastasis) can not be predicted, it has been suggested the term of well differentiated tumour of uncertain malignant potential. Finally it has to be mentioned the possible role of molecular biology in the diagnosis of well differentiated thyroid carcinoma; indeed markers such as RET/PTC or PAX8/PPARgamma, which to date have been employed mainly in basic research, might represent useful diagnostic (and therapeutic) tools in the future.