One of the principal reasons why speleothems are recognised as important palaeoclimate archives is their suitability for accurate and precise uranium-series (U-series) age determination. Sampling speleothem sections for U-series dating is straightforward in most cases because visible growth layers are preserved. However, this is not always the case, and here we describe a sampling strategy whereby growth layers are resolved from trace-element images produced by laser-ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). We apply this method to a section of an Italian subaqueous speleothem (CD3) that lacks persistent visible growth layering. Trace-element imaging revealed growth layers that are strongly non-planar in their geometry owing to the speleothem's pronounced euhedral crystal terminations. The most prominent trace-element layers were first digitized as x, y vector contours. We then interpolated these in the growth-axis direction to generate a series of contour lines at ∼250-μm increments. The coordinates of these contours were used to guide the sampling via a computerised micromilling lathe. This produced a total of 22 samples for U-series dating by multi-collector ICP-MS. The dating results returned ages in correct stratigraphic order within error. Close inspection of the U-series data and the derived depth-age model suggests that the main source of model-age uncertainty is unrelated to the contour sampling but instead more associated with how closely spaced the model ages are in time, i.e. the model age density. Comparisons between stable oxygen and carbon isotope profiles derived from aliquots of the dating samples and two other stable isotope profiles from CD3 spanning the same time period compare very favourably. Taken together, this suggests that our trace-element contouring method provides a reliable means for extracting samples for dating (and other geochemical analyses), and can be applied to similar speleothems lacking visible growth layering.