Classification of electrocorticography (ECoG)records related to finger movement is the mainpurpose of this study. Data set IV presented in BCICompetition IV was used in this paper. This data setcontains brain signals from three epileptic subjectsand the data records consist of both ECoG andelectronic glove data. ECoG segments related fingermovements were extracted by means of fingermovement records generated by electronic glove.Features of segments having different data pointswere extracted using autoregressive (AR) model.The AR coefficients were classified with SupportVector Machine (SVM) and K nearest neighbors(kNN) classifiers. AR coefficients were calculated using the leastsquares method. To get the acceptable result, the ARcoefficients were calculated in the range of [3-10]order. At the end of the analysis, the third-order ARcoefficients produces acceptable results wereobserved. k nearest neighbor (kNN) algorithm was introducedby Dasarathy in1991. It is a machine learningalgorithm that reads a set of labeled training set,and then it is used to classify an unlabeled testingset. In order to classify a testing pattern, it computesthe distance between testing pattern and all thetraining patterns. Then, the k training patterns ofclosest distance to the testing pattern are used todetermine the class of testing pattern. To get theacceptable results in our classification problem, thedistance of patterns were calculated with the kparameter in the range of [1-10] order. At the end ofthe analysis, generally the highest classificationperformances were observed with small k values.Support vector machine (SVM) is a discriminativeclassifier formally defined by a separatinghyperplane. In other words, SVMs are based on theconcept of decision planes that define decisionboundaries. A decision plane is one that separatesbetween a set of objects having different classmemberships. In our study, SVM was used to classifythe ECoG patterns with the kernel of radial basefunction. At the end of the SVM classification, the optimal values were obtained for regularization andsigma parameters of SVM. To find the finger which is hardest to classify amongall fingers, different applications were achieved. Inthe first application, it consists of two classes for P1and P5. In the second application, it consists ofthree classes for P1, P2 and P5. In the thirdapplication, it consists of four classes for P1, P2, P3and P5. In the last application, it consists of fiveclasses for P1, P2, P3, P4 and P5.The resultsshowed that performances were decreased withincreasing of classes, which was also an expectedresult. We listed the performances of SVM and kNNmethods. We found that the SVM method yielded thebest performance in classifying the ECoG patternrelated to finger movements. For the SVM classifier, the mean performances ofthree subjects were obtained as follows;classification rate 87.35% for two fingers,classification rate 66.97% for three fingers,classification rate 50.06% for four fingers andclassification rate 34.41% for five fingers.For the kNN classifier, the mean performances ofthree subjects were obtained as follows;classification rate 75.35% for two fingers,classification rate 55.50% for three fingers,classification rate 39.00% for four fingers andclassification rate 31.90% for five fingers.The results showed that we need differentapproaches to impove the performances. To get thehigher performance, more effective methods shouldbe improved. In the future work, we focused on thisproblem.