Structured-illumination microscopy delivers confocal-imaging capabilities and may be used for optical sectioning in bio-imaging applications. However, previous structured-illumination implementations are not capable of imaging molecular changes within highly scattering, biological samples in reflectance mode. Here, we present two advances which enable successful structured illumination reflectance microscopy to image molecular changes in epithelial tissue phantoms. First, we present the sine approximation algorithm to improve the ability to reconstruct the in-focus plane when the out-of-focus light is much greater in magnitude. We characterize the dependencies of this algorithm on phase step error, random noise and backscattered out-of-focus contributions. Second, we utilize a molecular-specific reflectance contrast agent based on gold nanoparticles to label disease-related biomarkers and increase the signal and signal-to-noise ratio (SNR) in structured illumination microscopy of biological tissue. Imaging results for multi-layer epithelial cell phantoms with optical properties characteristic of normal and cancerous tissue labeled with nanoparticles targeted against the epidermal growth factor receptor (EGFR) are presented. Structured illumination images reconstructed with the sine approximation algorithm compare favorably to those obtained with a standard confocal microscope; this new technique can be implemented in simple and small imaging platforms for future clinical studies.