To investigate the role of roughness in superhydrophobic coatings a variety of superhydrophobic and non-superhydrophobic surfaces were synthesized using various polymer binders, nanosilica particles and fluoro chemistry on both glass and polycarbonate substrates. The roughness of the coatings was measured by profilometry and atomic force microscopy (AFM) and analyzed by a variety of statistical methods. The skewness and kurtosis for the surfaces indicated that all coated samples, both superhydrophobic and non-superhydrophobic, had a random Gaussian roughness distribution. The skewness and kurtosis values, however, were not sufficient to determine superhydrophobicity, hence, other analysis methods had to be used to assess superhydrophobicity. It was concluded that the power spectral density function (PSDF) of the coatings was an effective tool to study the roughness needed to predict superhydrophobicity. Both RMS (root mean square) – the vertical scale of roughness and ACL (auto correlation length) – the horizontal scale of roughness that were extracted from the PSDF could predict superhydrophobicity. The ACL for all superhydrophobic formulations indicated that tightly packed asperities are essential for superhydrophobicity. In addition, the RMS roughness in nanoscale should be above 100 nm with ACL below 350 nm.