Weekly live cattle prices in various markets would be expectedto share common trends, i.e., they cannot be driven by separatenonstationarities because at some point the prices will divergesufficiently for it to be economic to cross-ship the cattle, or atleast the beef. This paper extends previous bivariate work to amultivariate analysis that is capable of modeling the multiplemarket linkages in prices for many geographical regions. Theempirical estimation represents the application of an innovation onAoki's Linear Systems State Space model that allows determination oflong- and short-run dynamics common to multiple series.The common dynamics permit characterization of the multiplemarkets with a limited number of states (sufficient statistics forthe past), resulting in dynamic arbitrage relations between theseries. A nonparametric test is used to evaluate the value ofexpected arbitrage forecasts implied by the structure of the model.The arbitrage relationship also is employed to generate efficientdiscounts/premiums for either physical delivery or cash settlement offutures contracts. The proposed settlement mechanism accounts forspatial arbitrage opportunities and therefore better represents thetrue geographical discounts faced by traders in individual markets.