Tree-ring studies have demonstrated that conifer latewood measurements contain information on long-term North American monsoon (NAM) variability, a hydroclimatic feature of great importance to plants, animals, and human society in the US Southwest. This paper explores data-treatment options for developing latewood chronologies aimed at NAM reconstruction. Archived wood samples for five Douglas-fir (Pseudotsuga menziesii, Mirb. Franco) sites in southeastern Arizona are augmented with new collections. The combined dataset is analyzed along with time series of regionally averaged observed precipitation to quantify the strength of regional precipitation signal in latewood time series and to identify ways of increasing the signal strength. Analysis addresses the signal strength influences of including or excluding “false” latewood bands in the nominal “latewood” portion of the ring, the necessary adjustment of latewood width for statistical dependence on antecedent earlywood width, and tree age. Results suggest that adjusted latewood width chronologies from individual sites can explain around 30% of the variance of regional summer (July–August) precipitation—increasing to more than 50% with use of multiple chronologies. This assessment is fairly insensitive to the treatment of false latewood bands (in intra-annual width and δ¹³C variables), and to whether latewood-width is adjusted for dependence on earlywood-width at the core or site level. Considerations for operational chronology development in future studies are (1) large tree-to-tree differences in moisture signal, (2) occasional nonlinearity in EW-LW dependence, and (3) extremely narrow and invariant latewood width in outer portions of some cores. A protocol for chronology development addressing these considerations is suggested.