We present a new proof of the monotonicity of the correction term theta(n) in Ramanujan's refinement of Stirling's formula. Moreover we prove that theta(n) is concave.