Limited attention has been paid to how well social mobility measures debated and used to study industrial countries perform in analysis of low-income settings. Following brief, selective reviews of the axiomatic and econometric literatures, three mobility concepts illustrate how properties that appear innocuous in industrial country analysis become problematic when downward mobility includes descents into destitution. For origin-independence measures-the most widely used in research on developing countries to date-axiomatic propriety and cognizance of co-residency-induced and other more well-known sources of estimation bias are not enough. This paper adopts the term 'perverse fluidity' from sociology to define the estimate bias attributable to intergenerational poverty descents. Using simple experiments and data from India, poverty descents generate perverse fluidity biases in intergenerational regression and correlation coefficients of up to 50 per cent, suggesting that seemingly 'good' mobility news may be 'bad' and that mobility comparisons are more precarious than acknowledged so far.