Matched-mode localization has been proposed recently as an alternative to matched-field processing for underwater source localization. It involves processing in mode space rather than phone space, and requires much less processing time compared to match-field processing. To date, studies of matched-mode processing have concentrated on single vertical arrays in range-independent environments. In this paper, matched-mode localization of a source in range, depth, and bearing using twin vertical arrays is demonstrated using simulated acoustic data. The environment considered consists of sloping bottoms with a sound speed increasing with depth. Acoustic modeling assumes adiabatic normal modes, neglecting horizontal refraction. Results of source localization are compared for both a range-independent environment and for matched-mode processing versus matched-field processing. It is found that for the cases studied the matched-mode processing is an order of magnitude faster than the matched-field processing for range-depth searches, and the time difference is greater for range-bearing searches. The matched-mode ambiguity surface shows equal or slightly improved (lower) sidelobe levels compared with that of matched-field processing. Results of source localization are compared for arrays of different number of phones, different vertical spacings, and different locations. The arrays that resolve more modes generally have better performance.