Although most known human teratogens often produce a combination of birth defects in an affected infant, surveillance programs aimed at detecting epidemics of birth defects usually only monitor rates of individual defects. A drawback to this approach is that an increase in the rate of infants affected with a specific combination of defects may lead to little or no increase in the rates of component defects. Using the Poisson distribution, we show that, compared with monitoring for individual defects, monitoring for combinations of two and three defects may require fewer numbers of births to detect an epidemic. In general, an increase can be detected more rapidly by monitoring the rates of defect combinations than by monitoring the rates of individual defects if most affected infants have combinations of defects rather than isolated defects. For example, in the case of Congenital rubella syndrome (CRS), monitoring for the combination of cataracts with deafness and/or patent ductus arteriosus could have led to earlier detection of an epidemic than could monitoring for cataracts alone. In contrast, in the case of thalidomide embryopathy, monitoring for reduction defects of upper limbs in combination with reduction defects of lower limbs and/or microtia/anotia would not have led to earlier detection of an epidemic than would monitoring for reduction defects of upper limbs alone. This is due mainly to the low frequency of defect combinations among affected cases. When used with regular monitoring for individual defects, surveillance of defect combinations can enhance the ability of monitoring programs to detect epidemics of birth defects.