Taylor’s power law (TPL hereafter) states that the variance of population size estimates is related to the mean population size by a simple power relationship, characterised by two coefficients, a and b. A long-term multi-species dataset on bird abundance was used to investigate the relationships between TPL coefficients, species demographic traits, habitat types, and also long-term trends in population dynamics. The relationships were investigated using a General Linear Model (GLM) framework. Comparison of TPL coefficients calculated from the first half of the long-term data  (1964-1982) and those calculated using the second half (1983-2000) showed surprisingly little correlation. There was also little correlation, within species, between b values calculated from plots occurring in different habitat types. These results raise concerns regarding the often-cited constancy of b. We show that the parameters of TPL can be related, to some degree, to life-history traits and we discuss the consequences of this for interpreting long-term spatio-temporal dynamics.