The input to online handwriting recognition system is themovement of digital pen on the (x, y) coordinates of a trace3. A stroke is the sequence of coordinates obtained fromone pen down to pen up 1.By studying the pen down to penup information of all the characters present in the databaseand based on the similarity they can be grouped into uniqueclasses. A character can be written either in a single stroke ora combination of multiple strokes. After the analysis of thedatabase, the maximum number of strokes required to writea character can be set 12 .Various classifiers like HiddenMarkov Model or HMM classifier 2,SVM 12,ARTMAPapproach 1,ANN,etc can be used as modeling techniqueto recognize the stroke.When an unknown character inputcomes up, the stroke classifier identifies a sequence ofstrokes which is then checked in reference set to recognizethe character.Various comparisions are also done in variousreasearch papers between various modeling techniques 12.The Hidden Markov Model or HMM classifier is a finitestate machine which have a set of hidden state 5.Each ofthe sample character contains within within it various strokeswhich are subdivided into substrokes.The feature vector of allthese substrokes are then converted into states.These statesform the state sequences.From the state sequences,the initialstate distribution and the state transition probabilities canbe estimated.HMM can solve the problem segmentation inpattern or character recognition 2.NN classsifier is based on Dynamic Time Wraping(DWT).It calculates the nearest prototype 6.It can matchtwo curves of unequal length which makes it an importantclassifier in online handwriting recognition.Another classifiercalled the MLP or the Multilayer Perception which is trainedby the BP or back propagated architecture is a neuralnetwork architecture used in handwriting recognition 6.Anotherneural network architecture based on Adaptive Resonancetheory(ART) is the fuzzy ARTMAP 1.It is capableof fast,stable,online,unsupervised or supervised,incrementallearning,classification and many more.The simplified fuzzyARTMAP can be developed by removing the redundanciesin the architecture.SFAM is faster than FAM(FuzzyARTMAP).It is a very useful technique in recognition ofonline handwriting of Indian script. Another classifier mostwidely used for recognition is SVM or Support Vector Machines.SVMs(Support Vector Machines)are techniques usedfor data classification.The main objective behind the SVMsis to produce a model based on the training data,whichestimates the target values of the test data.Polynomial kerneland Gaussian radial basis functions are the common kernelfunctions in SVM.