We use laboratory experiments and numerical models to quantify the effects of dike interaction on the focusing of magma as it ascends in the upper mantle. Laboratory experiments involve injecting buoyant fluid into the base of a tank filled with solidified gelatin. When we initiate two dikes parallel to each other, but separated by a horizontal distance x, they tend to merge as they ascend. This behavior is also predicted by numerical models of two-dimensional dikes. The key parameters that control the maximum horizontal separation xc over which dikes will intersect are dike driving pressures, dike head lengths L (i.e., the length over which driving pressure is large), and the difference between the principal stresses of the remote stress field. When the remote differential stress is small compared to the dike driving pressure, two dikes of equal driving pressure and length will intersect over distances of xc ~ L. This distance decreases with increasing remote differential stress. We quantify the effects on magma transport from a broad lateral distribution of magma using numerical simulations of multiple-dike interaction. When the average dike spacing prior to interaction is within ~3L and remote differential stresses are insignificant, dike interaction can focus magma over horizontal distances many times L but at least ~6L. Dike interaction can focus magma in the asthenosphere beneath mid-ocean ridges for low mantle viscosities (≤1019 Pa s) and if dikes initiate with average separations of a few hundred meters, or less. Such focusing is predicted to grow dikes of increasing magma flux approaching lateral separations of a kilometer.