There is now a substantial body of literature documenting the detectability of plants and animals under standard survey conditions. Despite the evidence that many flora and fauna species have detection probabilities of less than one, it is still the default assumption of most environmental impact assessment processes that if a species is present, it will be detected. Here we briefly review a number of existing studies that have estimated the survey effort necessary to detect animal species, based on what is known about their detection rates in standard surveys. We then propose a novel method, based on failure-time analysis, for quantifying the detectability of and determining appropriate survey effort for plant species during flora surveys. We provide computer code for implementing the method in the Bayesian freeware WinBUGS. Methods for estimating detectability can be used to inform minimum survey requirements and have important applications in environmental impact assessment and monitoring.