The current noise at the output of a microcalorimeter with a voltage biased superconducting transition edge thermometer is studied in detail. In addition to the two well-known noise sources: thermal fluctuation noise from the heat link to the bath and Johnson noise from the resistive thermometer, a third noise source strongly correlated with the steepness of the thermometer is required to fit the measured noise spectra. Thermal fluctuation noise, originating in the thermometer itself, fully explains the additional noise. A simple model provides quantitative agreement between the observed and calculated noise spectra for all bias points in the superconducting transition. © 2000 American Institute of Physics.