Efficient encoding of LPC parameters plays an essential role in low-rate speech coding. Frequently the principles of vector quantization (VQ) are applied to line-spectral-frequency parameters (LSF), which represent the short-term power spectrum of the speech signal. For complexity reasons, ordinary full-search VQ generally does not allow an adequate representation of the parameters, therefore suboptimal product VQ methods are often used instead. One realization is the split VQ proposed by Paliwal and Atal accomplishing transparent encoding of LSF parameters at 24 bits/frame [1],[2]. This paper presents a new approach to entropy coding (EC) of LSF parameters introducing a relative index coding scheme (RIC). Combined with various VQ procedures it substantially reduces the bitrate by exploiting correlations between successive frames. In application to ordinary VQ of LSF Parameters the new scheme yields bit savings up to 3.8 bits/frame. Combined with split VQ transparent quantization is achieved at approx. 17.5 bits/frame. As RIC works as a postprocessing scheme for the VQ indices no additional distortion arises.