High quality core samples recommended by testing standards or suggested methods for uniaxial compressive strength (UCS) determinations cannot always be obtained particularly from soft, weak-to-very weak and clay-bearing rocks. Due to this difficulty, some simple and cheap index test methods have been developed in order to indirectly estimate the UCS. However, preparation of smaller samples from such rocks even for some index tests is also difficult. In addition, sampling from historical sites, monuments and buildings for strength determinations in rock engineering studies is not allowed. To overcome these difficulties, a new and portable light-weight testing device, Needle Penetrometer (NP), has been developed in Japan. But there is no any standard or suggested method available for this test. It is a non-destructive test applicable both in the field and laboratory with minimum sample preparation. The use of NP was mainly concentrated on predicting the UCS of soft and weak-to-very weak rocks from the Needle Penetration Resistance (NPR) in Japan and more recently in Turkey by the authors, and a few empirical relationships between the UCS and NPR have been developed by some investigators. This most recent study aims to provide some new contributions to previous works on the NP test, to assess the performance of the test and to develop a more generalized empirical relationship for indirect estimation of the UCS from NPR based on the data obtained from a wider range of soft and weak-to-very weak rocks. In addition, the effect of penetration rate on test results was investigated and the possible uses of the NP test in geo-engineering with the recommendations of the authors were also briefly discussed. For these purposes, a database consisting of a total of 725 UCS-NPR data pairs from the previous studies of the authors and those from the investigators in Japan and additional new tests carried out in this most recent study was established. The experimental results suggested that the rate of penetration had no effect on the test results. This finding makes the use of NP advantageous both in the field and laboratory. Statistical analyses reveal that based on the database used, there is a significant relationship between UCS and NPR at the 95% confidence level and the UCS values predicted from the model recommended in this study show the best agreement with those experimentally determined when compared to those obtained from the existing models. Particularly 70% of the UCS predicted from this model fall into approximately +/- 7% scaled percent error. It was concluded that the NP tests can be applied to soft and weak-to-very weak rocks with UCS of up to 40 MPa to predict the UCS from NPR, however, its use on rock types consisting of coarse grains embedded in a cementing material, such as conglomerate, should be avoided. (c) 2012 Elsevier B.V. All rights reserved.