Substance use disorder (SUD) is a chronic psychopathology characterised by the persistent and uncontrolled use of drugs or alcohol (alcohol use disorder (AUD)) that places an enormous burden on affected individuals, as well as their families and society in general. In 2019, almost half a million people died due to drug use and harmful use of alcohol results in ~3 million deaths worldwide yearly (5.3% of all deaths). The large majority of SUDs and AUDs remain untreated, and when treatment is sought, this is often unsuccessful in controlling drug or alcohol intake. An intriguing phenomenon is that even after prolonged periods of abstinence, relapses to alcohol or drug use occur in more than half of the affected individuals. During drug or alcohol use, neutral environmental stimuli (cues) are associated with the rewarding effects of drugs or alcohol and thereby gain motivational value. These cues can trigger retrieval of drug- or alcohol-associated memories during abstinence, which induces craving and often results in relapse. It is therefore of pivotal importance to gain further insight into the neurobiological mechanisms that underlie cue-evoked relapse, to aid the development of effective pharmacotherapies. The persistent propensity of drug-associated cues to drive relapse indicates that drug use evokes long-term molecular and cellular neuroadaptations in brain regions that are involved in the processing and storage of cue-reward associations. The medial prefrontal cortex (mPFC) is considered one of the key brain regions in cue-evoked relapse to alcohol and cocaine use. Intriguingly, the mPFC can promote drug reward seeking, but also provide inhibitory control over drug seeking. Human neuroimaging studies revealed heightened activation of the mPFC upon cocaine- or alcohol-associated cues and the extent of activation predicts relapse risk. However, human studies do not allow further insight into the molecular and cellular processes involved in processing of cue-reward associations, which is why animal paradigms that model relapse to drug seeking have been developed. After associative learning between the drug and cues, re-exposing the animal to the drug- or alcohol-associated cues (in absence of reward) will evoke drug seeking behaviour, even after prolonged periods of abstinence. Previous studies using these models have revealed neuroadaptations in the mPFC that can contribute to cue-induced relapse. Moreover, it has been shown that learned associations are encoded by sparse, distributed populations of neurons, so-called neuronal ensembles. In my thesis I aimed to elucidate alcohol- and cocaine-associated memory processing in the mPFC at the level of cortical circuitry, neuronal ensembles and the proteome. Firstly, our data reveals that a neuronal circuit within the mPFC controls extinction learning through interaction between the vmPFC and dmPFC. This study further emphasises the important role of GABAergic PV interneurons in extinction of cocaine memory, as this interneuron population mediates the interaction between the vmPFC and dmPFC. Secondly, our study shows that an mPFC neuronal ensemble activated during alcohol SA functions as a long-lasting memory trace that can promote cue-driven relapse to alcohol seeking after prolonged abstinence. Furthermore, our study highlights that alcohol and natural reward associations are differently encoded by the brain. On the other hand, we report that alcohol re-intake in the home cage is not driven by an mPFC ensemble activated by home cage alcohol consumption. Finally, we show that molecular changes can be detected in the mPFC ensemble that stores a persistent alcohol-cue memory. This finding opens new avenues of research, as future studies should assess the causal contribution of these molecular changes in alcohol relapse and thereby may provide an important steppingstone for the development of novel therapeutic strategies to reduce relapse in AUD and cocaine use disorder.