Speech recognition occurs when attending to speech stimuli in auditory, visual, or audiovisual modalities under optimum (e.g., in silence) or degraded listening conditions (i.e., in background noise or in individuals with hearing impairment). The present thesis contains details of the first study to show how background noise (steady-state white noise) delayed the identification of different types of speech stimuli (consonants, words, and final words in sentence) in auditory and audiovisual modalities by calculating the isolation points (IPs, the shortest time from the onset of an speech stimulus required for correct identification of that speech stimulus). The cognitive demands of perceiving speech stimuli under each type of listening condition (silence vs. noise) were also tested by measuring the correlations between IPs of each type of speech stimulus (in each type of listening condition) and tests of working memory capacity (i.e., the reading span test, Rönnberg, Arlinger, Lyxell, & Kinnefors, 1989) and attentional capacity (i.e., the Paced Auditory Serial Attention Gronwall, 1977). In addition, the speechreading of words in both silent and noisy (babble background noise and steady-state white noise) conditions was also investigated in above-average speechreaders. The present thesis also compared elderly hearing-aid (EHA) users and elderly normal-hearing (ENH) individuals, in terms of IPs and accuracy in the identification of different types of speech stimuli in silence. Furthermore, the extent to which the identification of auditory speech stimuli was cognitively demanding (for both EHA users and ENH individuals) was investigated by measuring the correlations between IPs and accuracies of different types of speech stimuli and working memory capacity.The results showed that background noise resulted in later IPs (in the auditory and audiovisual modalities) and reduced accuracy (in the auditory, visual, and audiovisual modalities). In visual speech recognition (speechreading), the adverse effect of noise on accuracy was dependent on the type of background noise: babble background noise reduced accuracy but the steady-state white noise did not. In addition, the audiovisual presentation of speech stimuli, relative to auditory presentation, resulted in earlier IPs and more accurate identification of speech stimuli in both silence and noise. In the silent condition, the IPs for perceiving different types of auditory and audiovisual speech stimuli were not correlated with performance on the tests of working memory and attention. In the noise condition, the IPs of consonants and words in the auditory-only condition were significantly correlated with tests for measuring explicit cognitive resources. Furthermore, there was a significant correlation between the results on a test of speech-in-noise performance (the HINT, Hällgren, Larsby, & Arlinger, 2006) and the results on tests of working memory (the reading span test, Rönnberg et al., 1989) and attention (PASAT, Gronwall, 1977). A secondary finding was that ...