This paper introduces a novel approach for tool condition monitoring (TCM) in micromilling, which is based on the analysis of singularity of cutting force waveforms and the statistical distribution of the singularity measure. It is a multiscale approach with both localized and global cutting force analyses. The local singularity characteristics of the cutting force are measured by the Holder exponent at the point and determined by wavelet transform modulus maxima. The global characteristics are measured by the singularity spectrum, which represents the overall behavior of the singularity among different tool-wear states. Experimental studies from both micromilling of copper and steel show that the axial component of the cutting force is robust and most discriminant for TCM using its singularity measurements.