It is often taken for granted that measurement-level data fusion must necessarily give rise to an improvement in the track picture, particularly in situations where the sensors are spatially distributed. In many cases, this assumption is justified - an obvious example is where each sensor individually produces too sparse a set of plots to support a track, whereas multiple sensors can provide a sufficient density of measurements. Another example is where complementary viewing angles can permit more precise target localisation. There are, however, situations in which the provision of information from multiple sensors can actually be detrimental to the tracking of closely-spaced objects. This behaviour has been observed using very high-fidelity simulations of a set of high-range-resolution radars observing closely-spaced ballistic targets. It is shown that the complementary viewing angles provided by distributed sensors can actually increase the likelihood of miscorrelation, in situations in which the closely-spaced objects (CSOs) are resolvable in range but not in angle. An attempt to exploit the individual sensor resolution capabilities by using track-associated plot processing was found to be fragile. An MHT (Multi-Hypothesis Tracking) approach, however, is shown to be a robust solution.