Transmission spectroscopy of Earth-like exoplanets is a potential tool for habitability screening. Transiting planets are present-day Rosetta Stones for understanding extrasolar planets because they offer the possibility to characterize giant planet atmospheres and should provide an access to biomarkers in the atmospheres of Earth-like exoplanets, once they are detected. Using the Earth itself as a proxy we show the potential and limits of the transiting technique to detect biomarkers on an Earth-analog exoplanet in transit. We quantify the Earths cross section as a function of wavelength, and show the effect of each atmospheric species, aerosol, and Rayleigh scattering. Clouds do not significantly affect this picture because the opacity of the lower atmosphere from aerosol and Rayleigh losses dominates over cloud losses. We calculate the optimum signal-to-noise ratio for spectral features in the primary eclipse spectrum of an Earth-like exoplanet around a Sun-like star and also M stars, for a 6.5-m telescope in space. We find that the signal to noise values for all important spectral features are on the order of unity or less per transit - except for the closest stars - making it difficult to detect such features in one single transit, and implying that co-adding of many transits will be essential.