Species richness is the most widely used measure of biodiversity, but the relationship between the observed and true numbers of species present in a study site is not always investigated. A field study at 27 habitat remnants was used to measure the effectiveness of a survey regimen for assessing species richness of butterflies and day-active moths in southwest Western Australia. Observed species richness was compared with known species richness and to statistical estimates of true species richness, and the bootstrap was found to be the best predictor of true richness. A regimen of 10-m-wide walk transects sampled on six occasions at 2-wk intervals during the austral spring (mid-September to mid-December) gave an almost complete inventory of resident species for each site (≈87% of the fauna detected), consistent with two previous studies that have assessed sample completeness in temperate areas. The abundance of diurnal lepidoptera showed large temporal variation over the flight season and varied to a lesser extent with time of day and temperature, but not with cloud cover or wind speed. Transect route and sampling frequency were the most important considerations in devising a survey regimen: transects placed off tracks detected both more species and more individuals per unit length. The fraction of the site area sampled was relatively unimportant, and even low sampling fractions of 1–2% may be adequate if the number and frequency of surveys is sufficient. The design of future surveys would be facilitated if sampling fraction was routinely reported and examined in relation to sample completeness.