You can download either the 32-bit or 64-bit version of UCINET. The 32-bit version is the standard version and runs on both 32-bit and 64-bit Windows. The 64-bit version is experimental and only runs on 64-bit windows. As it is subject to crashes, it is best used in tandem with the 32-bit version.

32-bit Installation Package. This installs the 32-bit version of UCINET along with several helper programs (such as NetDraw), and puts a copy of all the standard datasets in a folder called Ucinet Data under your My Documents folder. The installation program is new as of version 6.531 and unfortunately can't automatically uninstall versions of UCINET prior to 6.531, so you should do that yourself prior to installing current version.

64-bit installation package. If you have a 64-bit version of Windows, you can try this 64-bit version of UCINET. It lets you analyze much larger datasets, if you have the RAM memory for it. However, it is also flaky. Older parts of UCINET are not compatible with 64-bit execution, so these parts will crash. If when you try to run the program it complains about missing some files, download and extract these files to the UCINET folder: Also, here is an (older) version of the 64-bit exe -- no installation package: UCI6.exe. If you download this one, you definitely need to download the file and extract the files to the same folder as uci6.exe

Trial vs "real" version. Actually, there is no separate trial version. If you download the program, it will run on your computer for 60 days without having to enter a registration code. Buying the program gets you that registration code. So whether you buy or just try, you download the same program.

Virus/malware issues. There was a time when the download files were stored in a DropBox location. For some reason, this caused Norton AntiVirus to flag the program as dangerous. So we don't use DropBox for this anymore. But you might still run into issues with Norton or some other anti-virus program when installing UCINET. One thing to be aware of is that Windows 8 will issue a warning that the publisher is unknown and probably dangerous. This just means we haven't sought (read: paid for) Microsoft certification.

Installation Notes


Installation works best if you right-click on the installation file (typically called something like setup32UCI6534.exe) and choose 'Run as Administrator'. Similarly, it is also helpful to use 'Run as Administrator' when running UCINET itself for the first time, as this allows you to register the program for all users of your machine. See the FAQ on this.


The best way to run UCINET on a Mac is to use a Windows emulator such as Parallels (or, of course, Bootcamp). However, it is (often) possible to run UCINET on a Mac using Wine. For more information, see this FAQ

Version Info

  • Version 6.570 | 22 May 2015
    • Small change in dialogue window for Network|Cohesion|E-I Index, adding "L" button to facilitate selection of attribute column
    Posted by Steve Borgatti
  • Version 6.569 | 18 May 2015
    Note: installation package updated 22 May to incude a new help file with proper support for CLI, and to include Netdraw 2.151. This fixes an important bug that now allows UCINET to invoke Netdraw with a 2-mode dataset. For example, in CLI you can now type:
    ->draw davis
    • Added two similar routines in CLI, ccsplit and ciasplit. Both partition a network into two groups based roughly on structural equivalence. CcSplit is the same as the first split obtained by running Networks|Roles & Positions|Structural|ConCorr. Optional parameters include minlev, which specifies the smallest (absolute) value of any correlation. CiaSplit is an experimental routine that is intended to give the same answers as ccsplit, but faster.
    Posted by Steve Borgatti
  • Version 6.568 | 16 May 2015
    Note: the installation package for 6.568 was updated 17 May 2015 to include additional data files
    • Added expected values as one of the outputs of Networks|Ego networks|Gould & Fernandez brokerage
    • Changed Networks|Centrality|Beta Centrality to allow calculation of either in-centrality (corresponding to in-degree) and out-centrality (like out-degree). In the past, it only computed out-centrality, requiring user to transpose matrix themselves
    • In CLI (aka matrix algebra), replaced inv and ginv functions with faster versions. Inv gives the inverse of a square, non-singular matrix. Ginv gives the Moore-Penrose generalized inverse, which can work with non-square matrices and can do a creditable job with singular or nearly singular matrices
    • Tools|Symmetrize now calculates a correlation coefficient between the original matrix and the symmetrized matrix in order to assess how much the matrix was changed
    • In Tools|Dichotomize Interactively, the variable "Separation" was renamed "Correlation". It gives the correlation between the original matrix and the newly dichotomized one.
    • Fixed bug in Matrix Editor in which the Fill function was defaulting to filling zero rows and columns.
    • Replaced Networks|Centrality|Information Centrality with faster routine
    • Added xtabs routine to CLI. Constructs contingency table of cross-classified frequencies for matrices or vectors. For example, to compare to relations, you might type:
      ->dsp xtabs(padgm padgb)   //padgm = marriage ties, padgb = business ties
      or to compare the results of two clustering routines (Split and NCD) you might type:
      ->dsp xtabs(split(zache) ncd(zache))  //zache is one of the networks in the Zackar datset
    Posted May 18, 2015, 6:20 AM by Steve Borgatti
  • Version 6.567 | 10 May 2015
    • Added measures of heterogeneity to Networks|Regions|Components|Binary
    Posted May 10, 2015, 5:23 PM by Steve Borgatti
  • Version 6.566 | 8 May 2015
    • Changed several dozen routines to respect the default output folder option, and also to use the current default folder when the input data is drawn from some other folder
    • Fixed bug in CLI version of Factions, which was having difficulties when the initial partition (drawn from NCD) had cluster id's bigger than the number of clusters. 
    • Added "Save Commands" button to CLI. This allows you to type a number of commands, and then save them all to a text file, which can then be used to rerun the commands. 
    • Added SocialProximity function to CLI which calculates, for each pair of nodes, the (number of friends in common plus 1) divided by the geodesic distance between them (where undefined distances are assigned a value of N). The idea is that one's social proximity to another is a function of both the extent they live in overlapping worlds, and the distance between them is short. If the number of friends in common is zero, then social proximity will be 1/dist. Note that in undirected graphs, it is not possible to have friends in common when the distance is greater than 2. For the directed case, the situation is more complicated. Example usage:
      ->sp = socialprox(campnet) 
    Posted May 8, 2015, 1:10 PM by Steve Borgatti
  • Version 6.565 | 4 May 2015

    • Changed CLI (aka matrix algebra) to accept "structuralholes" instead of just "holes", as in:

      ->h = structuralholes(campnet)

    • Fixed some recently introduced bugs in CLI with handling multiple arguments to functions. Beware that you may need to use full quotes around things that previously didn't need it. Still working on this.  
    • Replaced information centrality routine with something significantly faster
    • Added functions to Matrix Editor for copying row labels to columns and vice-versa
    Posted May 5, 2015, 11:18 AM by Steve Borgatti
  • Version 6.564 | 1 May 2015

    • Fixed bug in outputting expected values from P1 procedure
    • Attempted improvements to locating netdraw.exe and setting up temp folder on difficult windows installations
    • Installation package includes NetDraw 2.150, which includes a reorganized Layout menu
    Posted May 1, 2015, 11:21 AM by Steve Borgatti
  • Version 6.563 | 24 April 2015

    • Made internal changes to Beta Centrality to greatly increase speed
    • Also changed the default beta in Beta Centrality in cases were the principal eigenvector is complex. It now takes as the eigenvalue the sqrt of the square of the real part plus the square of the imaginary part.
    • Mode internal changes to Affiliations to slightly increase speed when calculating row-row network
    • Replaced Eigenvector Centrality with new version that is much faster
    • Added eigenvector routine to CLI. Syntax is Eigen(<dataset>), as in:
      ->x = eigen(padgm)
      the result is three files, x-leig (left eigenvector), x-reig (right eigenvector), and x-eig (eigenvalues). The routine still needs some work to handle complex values properly
    Posted Apr 25, 2015, 1:08 PM by Steve Borgatti
Showing posts 1 - 8 of 159. View more »