import arcpy arcpy.analysis.Buffer("landuse","landuse1000","1000 Meters")
import arcpy arcpy.analysis.Buffer("landuse","landuse1000",1)
I am assuming your data is based on just a Geographic Coordinate System and only uses decimal degrees for its coordinate values. If so, I believe you can only use angular units to work with that data. You would need to use the Project tool to convert it to a Projected Coordinate System that uses linear units such as meters and that is designed to project area correctly. Once you have chosen a good projection use the Project tool to transform your data into a new feature class you can buffer with linear units such as meters.
Melita Kennedy is the resident ESRI projection expert on the forum. Here is some advice she has given others on the subject of choosing an equal areas projection:
"You'll want to use an equal-area projection. You'll find some for the US under projected coordinate systems, continental, north america like "USA Contiguous Albers Equal Area Conic USGS". You could modify it and change the central meridian and standard parallels into your area of interest, but I'm not sure the census data is accurate enough to warrant that. When we've done testing of various equal area projections, the density of the data (how many vertices) had a greater effect that which equal area projection was used."
�?�If buffering a projected feature class that has features covering a large region, or you are using a very large buffer distance, distortions in the projection can cause inaccurate buffers to be produced. You can completely avoid distortion when buffering by using a feature class that has a geographic coordinate system and specifying a Buffer Distance in linear units (meters, feet, and so forth, as opposed to angular units such as degrees). When this combination of inputs is used, the tool will generate true geodesic buffers that accurately represent distances on Earth's surface. Geodesic buffers may appear unusual on a flat map, but when displayed on a globe these buffers will look correct (you can use the ArcGlobe or ArcGIS Explorer applications to view geographic data on a three-dimensional globe).