This paper proposes an effective method to compress SMS by doing some modifications to arithmetic coding data compression mechanism. The aim of this proposal is to optimize the maximum character capacity of SMS body. Every character in SMS is mostly encoded in 7 bit and the maximum capacity of oe SMS is only 1120 bit. Those SMS characteristic require a very efficient copression method to compress SMS.Arithmetic coding is a compression mechanism that works by converting a data message to a real code number between 0 and 1. Arithmetic coding provides nearly optimal data compression. However, it requires addtional memory space in compressed data to save arithmetic coding probability table for decompressing the copressed-data. Besides, it requires high precision and efective encoder-decoder to calculate and represent its code number (compressed-data). In very limited data space like SMS, the need of additional memory space to save arithmetic coding probability table is inefficient. It will make the compressed-SMS size bigger than the original SMS (uncompressed SMS) size. To overcome this inefficiency, in this paper, the need of memory space is erased. This paper proposes semi dynamic probability table usage to compress and decompress SMS for overcoming the inefficient need of memory space. To more optimize the effectiveness and efficiency of proposed-method compression ratio, this paper alsoo proposes a smart data representation to represent code numberso that he number of bits needed to represent compressed-SMS can be well-minimize. By using this smart data representation, 2k digit decimal code number value in base-10 can be written by only using k defalt GSM 7 it characters. The purposed compression mechanism in this paper has been researched plainly in mobile phone that uses android operating system. The SMS data test language used on the research is Bahasa Indonesia. Based on the research, the compression ratio of proposed compression mechanism is vary depends on thecontent of SMS. The average compression ratio of proposed compression mechanism is 71% while the maximum compression ratio is able to reach less than 25%, i.e. 500 character SMS can be compressed to 121 character SMS. Keywords: arithmetic coding, modification, SMS, compresion.