We used a geographic information system (GIS) to document spatial asso
ciations of Florida panthers, land cover, and other geographical featu
res. Panther radio locations (n = 14,548) occurred in hardwood hammock
: mixed hardwood swamp, and cypress swamp in greater proportion than i
n randomly positioned points (n = 8500). Panther radio locations occur
red less frequently in agricultural, barren, and shrub and brush land
cover. Panther home ranges consisted of a combination of preferred and
avoided cover types, including freshwater marsh, cypress swamp, hardw
ood swamp, and agricultural land. These cover types accounted for 62%
of the area in panther home ranges. We used correlation and discrimina
nt function analyses to assess the potential importance of 20 landscap
e features. These panther locations were effectively distinguished fro
m random points using four landscape variables: (1) the size of a cont
iguous patch of preferred land cover; (2) the proximity to preferred l
and cover; (3) the diversity of three preferred cover types within a w
indow 120 X 120 m, and (4) the matrix within which preferred cover typ
es occurred Eighty-three percent of the panther locations and 81.9% of
the random points were correctly classified based on a linear model c
onstructed using these four variables. Large, contiguous areas of pref
erred land-cover types were especially important because 96% of all pa
nther locations occurred within 90 m of preferred land cover. The aver
age preferred forest patch size that was used by these panthers was 20
,816 ha, and a regression equation suggests that patches larger than 5
00 ha are important. Maps of panther habitat suitability were develope
d using coefficients derived from discriminant analysis. Large areas o
f suitable land cover that are heavily used by panthers occur on priva
te ranches covering 3606 km(2). Conservation of preferred habitat on t
hese private lands is essential to maintaining a free-ranging populati
on of panthers in southwest Florida.