Liquid biopsy-derived circulating tumour cells (CTCs) could greatly potentiate precision oncology by yielding actionable biomarkers spanning molecular (oncogenic/resistance) signals to cancer cell-phenotypes; all via minimally invasive sampling that is compatible with longitudinal patient monitoring (unlike solid biopsy). Yet typical CTC profiling via standard 4-5 marker immunofluorescence imaging provides insufficient molecular bandwidth to guide precision oncology given adaptable, diversifying disease-states and an ever-growing therapeutic arsenal. To achieve a step-change in CTC utility as molecular guides for precision oncology, we have increased CTC profiling depth at least ten-fold via an end-to-end pipeline for deep multiplexed imaging, capturing 50 molecular markers per CTC. We quantify the expression, phosphorylation and subcellular localization of known and putative biomarkers that: define CTCs and cancer phenotypes; directly read-out oncogenic/resistance signalling, and/or; are explicit therapeutic targets. Validated via detection of known resistance-biomarkers in a three-stage resistance-progression model of prostate cancer, translational analysis of prostate cancer patient-derived CTCs then confirmed that our deep multiplexed profiling: i) improves CTC classification and; ii) captures inter- and intra-patient CTC heterogeneity corresponding to therapy responses. Machine learning then identified; iii) therapeutically actionable profiles per patient, integrating molecular expression and subcellular localization. This demonstrates proof-of-principle capacity for deep multiplexed CTC image-profiling to derive unprecedented molecular and cellular insights suited to guiding precision oncology.