Research in the field of language treatment and rehabilitation to date has used single-word naming as a controlled measure of outcome. Yet, given people do not actually communicate in single words, there is much debate as to whether this approach to assessment actually captures or measures real-world communication difficulties. This article aimed to add to this discussion by conducting a review of the literature on the relationship between picture-naming accuracy and word retrieval in connected speech. A search of PubMed, PsychInfo and Google Scholar identified 997 results; after exclusion of duplicates, screening of abstracts and full texts and screening of references, 11 articles remained. Seven of the 11 articles used standardised tests to measure picture naming, whist the remaining four developed their own. Five of the 11 used personal narrative to generate connected speech, three used story retelling, one used a picture sequence, one used procedural narrative, one video narration and one conversation. Most commonly, these were assessed based on their informativeness. There was no consistent correlation between naming and connected speech in these studies. These studies used inconsistency in the methods used across studies and future research needs to identify a core set of connected speech measures, explore the correlation between these and naming, and investigate the impact of participant characteristics on these. Given the need for rigour and ecological validity in research on complex interventions, this preliminary work is essential in moving research forward in this field.