The morphological variability (coiling properties, size and shape) of the planktic foraminifer Contusortuncana contusa (Cushman) in the terminal Cretaceous ocean was examined at eight deep-sea sites and two continental sections from low (16?) to middle (42?) paleolatitudes in both hemispheres. The material used in this study includes samples from the South Atlantic (DSDP Sites 356, 527 and 525A), North Atlantic (Sites 384 and 548A), Indian and Pacific Oceans (DSDP Site 465A and ODP Sites 761C and 762C) and Tethyan Ocean (outcrop sections from El-Kef and Caravaca). On average 45 specimens from two samples per location were analysed, from an interval corresponding approximately to the last 60 kyr of the Cretaceous. No differences in coiling direction (dextral proportions were > 90% in all samples), percentage of kummerform specimens (usually > 50%) and number of chambers in the last whorl (4-5) were observed between the sites. Both test size (expressed as spiral outline area and test volume) and total number of chambers increase significantly towards lower latitudes. Similarly, test conicity, examined by shape coordinate and eigenshape methods, and angularity of the spiral outline show a rather continuous, slight increase towards lower latitudes. Kummerform specimens of C. contusa were slightly larger and more conical than normalforms and possessed substantially more chambers (both totally and in the last whorl). A principal components analysis of the sample means of five variables describing size and shape clearly distinguished high-latitude sites (525A, 527, 548A, 761C and 762C) from low-latitude sites (384, 465A, Caravaca and El-Kef). Specimens from Site 356 are transitional with respect to those two groups. The results indicate: (1) considerable morphological variation in C. contusa during the terminal Cretaceous comparable to that known in many Recent planktic foraminiferal species and (2) a geographical distribution of this variation corresponding to previously suggested biogeographic schemes based on quantitative analysis of planktic foraminiferal assemblages. Despite the differences in sample means, the overall morphology of C. contusa overlaps among the sites studied, supporting the classification of all C. contusa morphotypes as a single species. Similarly, no discrete morphologic groups could be distinguished within any of the samples.