Abstract: The goal of this study was to investigate the effects of noise from building mechanical systems with time-varying fluctuations on human task performance and perception, and to determine how well current indoor noise rating methods account for this performance and perception. Six different noise conditions with varying degrees of time-varying fluctuations, many focused in the low frequency rumble region, were reproduced in an office-like setting. Thirty participants were asked to complete typing, grammatical reasoning, and math tasks plus subjective questionnaires, while being exposed for approximately one hour to each noise condition. Results show that the noise conditions with higher sound levels (greater than 50 dBA) combined with excessive low frequency rumble as well as those with larger timescale fluctuations (i.e., a heat pump cycling on and off every 30 seconds) were generally perceived to be more annoying than the other signals tested, although statistically significant negative relationships to task performance were not found. Other findings are (1) that the noise characteristics most closely correlated to higher annoyance/ distraction responses in this study were higher ratings of loudness followed by roar, rumble, and changes in time; and (2) that perception of more low frequency rumble in particular was significantly linked to reduced performance on cognitively demanding tasks. As for the ability of current indoor noise rating systems to match human performance or perception, none of the indoor noise rating methods evaluated were significantly correlated to task performance, but aspects of subjective perception such as loudness ratings were statistically related. Spectral quality ratings included with some noise rating methodologies were inconsistent with subjective perception, but other metrics such as RNC, L1 – L99 [LF ave], and LCeq – LAeq, were strongly correlated to rumble perception. The authors use the results to suggest a framework for an ‘ideal’ indoor noise rating method, but further research is required towards quantifying specific guidelines for acceptable degrees of time-varying fluctuations and tonalness.