Background: In addition to the triad of symptoms required for a diagnosis of autism spectrum disorder (ASD), individuals with ASD also commonly display abnormal sensory sensitivity (e.g. hypersensitivity to acoustic stimuli), although the underlying mechanisms responsible for this phenomenon remain unclear at present. In particular, parental and self-report data suggest that individuals with ASD experience a greater-than-normal difficulty in the perception of speech in the presence of interfering background sounds, be they noise or competing speech. Objectives: To conduct a series of studies designed to investigate the mechanisms underlying speech-in-noise perception in ASD individuals. Methods: Adaptive psychophysical procedures were employed to assess: (1) the detection of sinusoidally amplitude-modulated noise carriers for a range of modulation frequencies (2-512 Hz); and (2) speech-in-noise perception (i.e. speech reception thresholds (SRT), defined as the signal-to-noise ratio required for 50% speech intelligibility), using a range of background speech-shaped sounds, including a steady-state noise, a single competing talker, a temporally modulated noise, a spectrally modulated noise, and a spectro-temporally modulated noise. Three age- and IQ-matched normal-hearing groups were used: high-functioning individuals with autism (HFA); individuals with Asperger's syndrome (AS); and typically-developing control participants. Results: Individuals with HFA and AS had higher (i.e. poorer) modulation detection thresholds than control participants at all modulation frequencies tested. In addition, both groups displayed poorer SRTs than the controls, but only for background sounds that contained temporal modulations in amplitude. Conclusions: Individuals with ASD appear to be less capable of extracting target speech signals present in regions where the background contains energy minima, than the control participants. This does not seem to be due to a deficit in temporal resolution (i.e. fast temporal processing deficit), but rather as a result of poorer-than-normal temporal processing efficiency. No significant differences in performance were found for the HFA and AS groups.