Résumé : Abstract Paraffin sections from animal or human tissues fixed in different fixatives were submitted to immunostaining with the mouse monoclonal antibody 19A2, developed by Ogata et al. (1987a) against cyclin/PCNA. Detection of the bound antibody was performed by the indirect method with biotinylated sheep antibody and streptavidin‐biotin‐peroxidase complexes. No, or faint, nuclear staining was seen in material fixed in ethanol, Bouin, Bouin‐Hollande, Carnoy or formaldehyde, whereas readily detectable immunocytochemical reaction was constantly observed over nuclei of methanol‐fixed tissues. Hydrolysis with 2 N HCl prior to immunocytochemistry (as currently performed to render incorporated BrdU accessible to antibodies) somewhat improved the results with Bouin or Carnoy and markedly augmented the intensity of the peroxidase reactions in formaldehyde and in methanol‐fixed tissues. The distribution of the positive nuclei in the two latter cases coincided with the proliferative compartment. On the other hand, double labelling with [3H]‐thymidine and with the cyclin/PCNA antibody revealed that in methanol‐fixed tissues the cyclin/PCNA labelling index did not differ by more than 6% from the [3H]‐thymidine index. Besides the two labels overlapped in a proportion of labelled cells that was in reasonable agreement with expectation considering cells flow in and out of S phase since the time of [3H]‐thymidine injection. This indicates that both labels recognize the same cells in this material. In contrast, in formaldehyde‐fixed tissues, the cyclin/PCNA labelling index markedly exceeded the [3H]‐thymidine labelling index. From this it is concluded that cyclin/PCNA immunostaining can be used:  In formaldehyde‐fixed tissues (including existing material stored as paraffin blocks): for defining and mapping the proliferative (or germinative) compartment.  In methanol‐fixed tissues as a substitute to the [3H]‐thymidine autoradiographic labelling index. From this, a method is proposed (derived from classical ‘double‐labelling’technique) for measuring S phase duration in tissues fixed at a known interval time after a single labelling with [3H]‐thymidine (or BrdU) and submitted to cyclin/PCNA immunocytochemical detection and to autoradiography (or to BrdU immunostaining). Copyright © 1989, Wiley Blackwell. All rights reserved