When an individual perceives a situation or stimulus as anxiety-provoking they may react behaviourally; often actions are carried out that make it possible for the individual to cope with the anxiety. Thus, the individual comes to associate the elicited behaviour with a decrease in anxiety. Potentially, when such behaviours are carried out, conditioned inhibitors, or safety signals, are generated. On theoretical grounds, these are expected to help maintain and secondarily reinforce the behaviour. The current thesis examined both conditioned inhibition and the learning of stimulus–response associations in both a healthy sample and a clinical sample of participants with anxiety disorder and/or problems with substance abuse. Two novel tasks were developed and one previously used task was used to examine conditioned inhibition, Negative Images CI Task: Retardation Test, Negative Images CI Task: Summation Test and ‘Mission to Mars’ CI Task: Summation test respectively. Four response inhibition tasks were developed to examine any accuracy or reaction time differences to neutral and emotional stimuli: Emotional Stroop Task, Go/No-Go Words Task, Go/No-Go OCD Colour Images Task, Go/No-Go Black and White Images Task. Performance on all of the tasks was correlated with individual differences in anxiety as measured by questionnaires: HADS, MOCI, BIS/BAS and the EPQR-S. The results from the healthy sample tested showed clear evidence of discrimination learning, as well as conditioned inhibition as measured by both retardation and summation tests. There were also response inhibition differences on the Emotional Stroop, a classic Stroop effect, less accurate and slower for colour incongruent words compared to other word-types, and more accurate and quicker responses to negative and OCD related words. There were no response inhibition differences on any of the Go/No-Go tasks. Further to this, in general, individual differences in anxiety as measured by the HADS, MOCI, BIS/BAS and EPQR-S were related to performance on the tasks. The hypothesis was that individuals formally diagnosed with an anxiety disorder would show better conditioned inhibition and response inhibition deficits. Recruitment for the clinical sample was unexpectedly difficult and therefore the sample size provides only preliminary data. The results from the clinical sample tested showed no difference in performance on any of the tasks; thus a formal clinical diagnosis of either an anxiety disorder or substance abuse disorder did not measurably impact on performance. However, overall the clinical group did not show discrimination learning or conditioned inhibition. On the Emotional Stroop Task the clinical sample showed a classic Stroop effect for accuracy and was also more accurate for negative words but there was no difference in latencies. There were no differences in performance on any of the Go/No-Go tasks. Across all of the tasks the clinical sample demonstrated a relationship between task stimuli and individual differences as measured by the HADS, MOCI, BIS/BAS and EPQR-S related to performance. The results from the current tasks demonstrated that inhibitory processes are influenced or affected by individual differences in anxiety in a healthy sample; in particular certain measures either positively or negatively influence performance. In order for this to be fully conclusive all of the tasks carried out need to be tested in a larger clinical sample. The results have implications for psychological treatments, for example, cognitive behavioural therapy (CBT). CBT is based on associative learning principles, if safety signals were identified in the maintenance of the anxiety these could be incorporated into therapy and aid the breakdown of negative associations formed.