A survey among natural rubber importers in France has shown that natural rubber from African producing countries appears to be more variable than that from Asia. In order to overcome this problem and check the observations reported, an initial study on the variability of certain natural rubber properties, and on sampling problems, was covered in a report issued in November 1998 for 5 factories: two in Côte d'lvoire (SOGB and SAPH), one in Cameroon (CDC), one in Gabon (HEVEGAB) and one in Nigeria (PAMOL). This initial study on the variability of African natural rubber (five factories involved) provided a picture at a given moment of the dispersion of physico-chemical characteristics within natural rubber bales on the production lines in the African factories taking part in the study. The protocol used took into account a large number of samples per production line, for which only the conventional quality criteria (ISO2000) were measured. This study revealed large disparities in the average level, but also in the dispersion of certain properties. With the resumption of the CFC project in June 2000, under the control of UNOPS and IRSG, a second protocol on the measurement of African natural rubber variability was scheduled. However, this protocol had to undergo some minor modifications during its implementation in 2002. Only 4 out of the 6 factories originally planned took part in the study. The main aims of this new study were to: - confirm or refute the results obtained in 1998 for the production units in the producing countries involved (Côte d'lvoire, Cameroon, Nigeria and Gabon), - introduce new physico-chemical criteria (average molar masses, gel contents, fatty acid index, etc.) to qualify such variability more effectively, - integrate observations and criteria specific to manufacturers, to ensure the validity of the specifications for SAR grade rubber, and its acceptability on the export market. This approach should enable identification of the causes of the va