The relationship between prevalence, incidence, and duration of disease is studied in exponentially growing/declining stable populations. Prevalence odds is shown to be a weighted average of age-specific products between incidence and discounted disease duration. If and only if the covariance between incidence and duration is zero, does prevalence odds equal the product of average incidence and average duration. The product of averages is shown typically to overestimate prevalence in epidemiologic applications. Ignoring population growth also tends to lead to overestimation of prevalence.