Abstract. Attempts to monitor the genetic variation of endangered populations by the use of blood protein electrophoresis often suffer from three drawbacks: a small sample of loci, lack of control populations with “normal” variation, and, sometimes, difficulty in confirming inheritance of electromorphs. An endangered isolate (Hamilton, Victoria) of the eastern barred bandicoot, Perameles gunnii, was compared with a widespread, dense, conspecific population in Tasmania. A previous study of the effective size of the siolate suggested that the loss of variation should be detectable by protein electrophoresis if average heterozygosity (h̄) was approximately 0.057 in widespread, dense populations and 20 to 50 individuals from each population were analysed for 25 or more loci. However, no genetic variation was detected within or between samples. Similar studies proposed as a baseline for monitioring genetic variation could be equally powerless to detect changes in variation, even with quite high h̄ values. The analysis of variation in DNA is expected to avoid many of the problems associated with blood protein studies. This study highlights the importance of a control population. Analysis of the Hamilton population alone might have led us to concludce that the recent population crash has been responsible for the low variation; however, this conclusion is not warranted, because h̄= 0 in the much larger Tasmania population.