One aspect of the Global Positioning System (GPS) is the potential to conduct geophysical research, and worldwide networks of GPS receivers have been established to exploit this potential. Several research groups have begun using this global GPS data to study ionospheric total electron content (TEC) variations, also referred to as GPS phase fluctuations, as surrogates for ionospheric scintillations. This paper investigates the relationship between GPS amplitude scintillations and TEC variations for the same line of sight using observations from Anc¿n, Peru. These observations were taken under equatorial spread F conditions for three nights in April 1997. As expected, only when the spectrum of TEC fluctuations includes significant power at the Fresnel scale do scintillations appear. We also find that when the TEC fluctuation spectrum includes the Fresnel scale, the S4 scintillation index is roughly proportional to measures of TEC fluctuation for the weak scintillations observed. The proportionality constant varies from night to night, however, casting doubt on the ability to predict GPS S4 successfully from TEC fluctuation data alone. We also present a simple theoretical phase screen model and show that if a relationship between TEC fluctuation measures and S4 exists, that relationship depends on the power spectrum of phase variations at the screen. Unfortunately, the available TEC data, at 30 s per sample (with some aliasing apparently permitted), offer limited spectral information. A preliminary comparison of 1 s/sample data with the same data decimated to a 30 s/sample interval suggests, however, that the level of successful S4 prediction, based on TEC fluctuation measures alone, is comparable at either sample rate. ¿ 1999 American Geophysical Union