Neutral landscape models (NLM) can provide standards against real landscapes, and are used to describe landscape pattern and process in the past few decades. In this paper, the neutral landscape models RULE and SimMap were tested against a real landscape, and a set of landscape metrics were used to quantify the spatial characteristics of real and simulated patterns. Measurements of some metrics (total number of patches, total perimeter, average patch area, aggregation index, contagion and lacunarity) suggested that definite level of consistency between NLM-generated maps and real landscape did exist at landscape or class levels. But, there were also some metrics, such as corrected patch perimeter-area ratio, fractal double-logged and edge distribution evenness, which didn't show any agreement between the generated maps and real landscape. In all, each NLM had its own strength in representing real landscape, but none of them was perfect.