Previous research on 'denied pregnancy', i.e. lack of subjective awareness of pregnancy until the end of gestation in pregnant women, is reviewed and reinterpreted in an evolutionary biological framework. Recent epidemiological studies show that this condition has a much higher incidence than previously thought (about 1:475). Very often, bodily symptoms of pregnancy (nausea, amenorrhea and abdomen swelling) are absent or greatly reduced, and neonates tend to be underweight; in many cases, pregnancy goes undetected also by relatives and physicians. Current explanations in the clinical literature are based on psychodynamic hypotheses about pregnancy-related unconscious conflicts; the lack of symptoms is accounted for by 'somatic denial'. I argue that such psychodynamic accounts are misguided for two reasons: (1) they rest on a failure to recognize the active biological role of the fetus in determining the course of pregnancy, and (2) they ignore the many levels of mother-fetus conflict over resource allocation described by biological theories of parent-offspring conflict. Here I propose to redefine this condition as 'cryptic pregnancy', and begin to explore its possible physiological correlates and evolutionary significance. In the light of parent-offspring conflict theory, cryptic pregnancy appears to reduce the costs of pregnancy, both energetic and ecological (mobility, dependence on kin/mate, etc.), thus favoring the mother at the expense of the fetus. Reduced hCG production and/or effectiveness is likely to be involved in the process. I propose and discuss three nonexclusive evolutionary hypotheses to account for this phenomenon: (1) cryptic pregnancy could be a nonadaptive outcome of conflict resolution processes over resource allocation in pregnancy, possibly related to minor disruptions of genomic imprinting mechanisms. (2) Cryptic pregnancy could result from missed spontaneous abortions of low-quality fetuses. (3) Finally, cryptic pregnancy could be an adaptive pattern of 'forced cooperation' between mother and fetus in stressful or threatening ecological circumstances, as suggested by the reported association with elevated psychosocial stress. In case of reduced survival probability, both mother and fetus would benefit if the mother reduced investment in pregnancy in order to maximize her chances of surviving and reaching delivery.