We evaluated wildlife population health from the perspective of statistical means vs. variances. We outlined the choices necessary to provide the framework for our study. These consisted of spatial and temporal boundaries (e.g., choice of sentinel species, populations, time frame), measurement techniques (molecular to population level), and appropriate statistical analyses. We chose to assess the health of 19 sea otter populations, located in the north Pacific from the Aleutian Islands, AK, to Santa Barbara, CA, and varying in population growth rates and length of occupancy. Our focal metric was gene expression (i.e., mRNA transcripts) data that we had previously generated across sea otter populations as a measure of population health. We used statistical methods with different approaches (i.e., means vs. variances) and examined the subsequent interpretive outcomes and how these influence our assessment of “health.” Interpretations based on analyses using variances versus means overlapped to some degree. In general, sea otter populations with low variation in gene expression were limited by food resources and at or near carrying capacity. In populations where the variation in gene expression was moderate or high, four out of five populations were increasing in abundance, or had been recently increasing. Where we had additional information on sources of stressors at the level of the population, we were able to draw inferences from those stressors to specific gene expression results. For example, gene expression patterns of sea otters from Western Prince William Sound were consistent with long term exposure to petroleum hydrocarbons, whereas in Kachemak Bay, patterns were consistent with exposure to algal toxins. Ultimately, determination of population or ecosystem health will be most informative when multiple metrics are examined across disciplines in the context of specific scenarios and goals.